End of training
Browse files- all_results.json +11 -0
- eval_results.json +7 -0
- pytorch_model.bin +1 -1
- train_results.json +7 -0
- trainer_state.json +3949 -0
all_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_loss": 0.15362989902496338,
|
| 4 |
+
"eval_runtime": 100.2898,
|
| 5 |
+
"eval_samples_per_second": 89.152,
|
| 6 |
+
"eval_steps_per_second": 1.396,
|
| 7 |
+
"train_loss": 0.07169675377377292,
|
| 8 |
+
"train_runtime": 490800.9612,
|
| 9 |
+
"train_samples_per_second": 33.299,
|
| 10 |
+
"train_steps_per_second": 0.13
|
| 11 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_loss": 0.15362989902496338,
|
| 4 |
+
"eval_runtime": 100.2898,
|
| 5 |
+
"eval_samples_per_second": 89.152,
|
| 6 |
+
"eval_steps_per_second": 1.396
|
| 7 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 967095881
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8a1d1d623f212485240ab30a7eb4624af743541fbf2b90920b2be7fc6ca7c504
|
| 3 |
size 967095881
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 0.07169675377377292,
|
| 4 |
+
"train_runtime": 490800.9612,
|
| 5 |
+
"train_samples_per_second": 33.299,
|
| 6 |
+
"train_steps_per_second": 0.13
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3949 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.15402325987815857,
|
| 3 |
+
"best_model_checkpoint": "whisper-small-full-data-language-v2-20ep/checkpoint-60000",
|
| 4 |
+
"epoch": 20.003133322888925,
|
| 5 |
+
"global_step": 63840,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.03,
|
| 12 |
+
"learning_rate": 2.5000000000000004e-07,
|
| 13 |
+
"loss": 2.0939,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.06,
|
| 18 |
+
"learning_rate": 5.000000000000001e-07,
|
| 19 |
+
"loss": 1.2605,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.09,
|
| 24 |
+
"learning_rate": 7.5e-07,
|
| 25 |
+
"loss": 0.5371,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.13,
|
| 30 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 31 |
+
"loss": 0.3785,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.16,
|
| 36 |
+
"learning_rate": 1.25e-06,
|
| 37 |
+
"loss": 0.3296,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.19,
|
| 42 |
+
"learning_rate": 1.5e-06,
|
| 43 |
+
"loss": 0.3046,
|
| 44 |
+
"step": 600
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.22,
|
| 48 |
+
"learning_rate": 1.7500000000000002e-06,
|
| 49 |
+
"loss": 0.2878,
|
| 50 |
+
"step": 700
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.25,
|
| 54 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 55 |
+
"loss": 0.2725,
|
| 56 |
+
"step": 800
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.28,
|
| 60 |
+
"learning_rate": 2.25e-06,
|
| 61 |
+
"loss": 0.2628,
|
| 62 |
+
"step": 900
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.31,
|
| 66 |
+
"learning_rate": 2.5e-06,
|
| 67 |
+
"loss": 0.2515,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.34,
|
| 72 |
+
"learning_rate": 2.7500000000000004e-06,
|
| 73 |
+
"loss": 0.2439,
|
| 74 |
+
"step": 1100
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.38,
|
| 78 |
+
"learning_rate": 3e-06,
|
| 79 |
+
"loss": 0.2333,
|
| 80 |
+
"step": 1200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.41,
|
| 84 |
+
"learning_rate": 3.2500000000000002e-06,
|
| 85 |
+
"loss": 0.2262,
|
| 86 |
+
"step": 1300
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.44,
|
| 90 |
+
"learning_rate": 3.5000000000000004e-06,
|
| 91 |
+
"loss": 0.222,
|
| 92 |
+
"step": 1400
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.47,
|
| 96 |
+
"learning_rate": 3.75e-06,
|
| 97 |
+
"loss": 0.2162,
|
| 98 |
+
"step": 1500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.5,
|
| 102 |
+
"learning_rate": 4.000000000000001e-06,
|
| 103 |
+
"loss": 0.2122,
|
| 104 |
+
"step": 1600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.53,
|
| 108 |
+
"learning_rate": 4.250000000000001e-06,
|
| 109 |
+
"loss": 0.2074,
|
| 110 |
+
"step": 1700
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.56,
|
| 114 |
+
"learning_rate": 4.5e-06,
|
| 115 |
+
"loss": 0.2034,
|
| 116 |
+
"step": 1800
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.6,
|
| 120 |
+
"learning_rate": 4.75e-06,
|
| 121 |
+
"loss": 0.1998,
|
| 122 |
+
"step": 1900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.63,
|
| 126 |
+
"learning_rate": 5e-06,
|
| 127 |
+
"loss": 0.1961,
|
| 128 |
+
"step": 2000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.66,
|
| 132 |
+
"learning_rate": 5.25e-06,
|
| 133 |
+
"loss": 0.1892,
|
| 134 |
+
"step": 2100
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.69,
|
| 138 |
+
"learning_rate": 5.500000000000001e-06,
|
| 139 |
+
"loss": 0.1873,
|
| 140 |
+
"step": 2200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.72,
|
| 144 |
+
"learning_rate": 5.750000000000001e-06,
|
| 145 |
+
"loss": 0.1843,
|
| 146 |
+
"step": 2300
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.75,
|
| 150 |
+
"learning_rate": 6e-06,
|
| 151 |
+
"loss": 0.1881,
|
| 152 |
+
"step": 2400
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.78,
|
| 156 |
+
"learning_rate": 6.25e-06,
|
| 157 |
+
"loss": 0.1813,
|
| 158 |
+
"step": 2500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.81,
|
| 162 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 163 |
+
"loss": 0.1828,
|
| 164 |
+
"step": 2600
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.85,
|
| 168 |
+
"learning_rate": 6.750000000000001e-06,
|
| 169 |
+
"loss": 0.1782,
|
| 170 |
+
"step": 2700
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.88,
|
| 174 |
+
"learning_rate": 7.000000000000001e-06,
|
| 175 |
+
"loss": 0.1785,
|
| 176 |
+
"step": 2800
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.91,
|
| 180 |
+
"learning_rate": 7.25e-06,
|
| 181 |
+
"loss": 0.1725,
|
| 182 |
+
"step": 2900
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.94,
|
| 186 |
+
"learning_rate": 7.5e-06,
|
| 187 |
+
"loss": 0.1713,
|
| 188 |
+
"step": 3000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.97,
|
| 192 |
+
"learning_rate": 7.75e-06,
|
| 193 |
+
"loss": 0.1684,
|
| 194 |
+
"step": 3100
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.0,
|
| 198 |
+
"learning_rate": 8.000000000000001e-06,
|
| 199 |
+
"loss": 0.1694,
|
| 200 |
+
"step": 3200
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.03,
|
| 204 |
+
"learning_rate": 8.25e-06,
|
| 205 |
+
"loss": 0.1642,
|
| 206 |
+
"step": 3300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.07,
|
| 210 |
+
"learning_rate": 8.500000000000002e-06,
|
| 211 |
+
"loss": 0.1608,
|
| 212 |
+
"step": 3400
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.1,
|
| 216 |
+
"learning_rate": 8.75e-06,
|
| 217 |
+
"loss": 0.1578,
|
| 218 |
+
"step": 3500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.13,
|
| 222 |
+
"learning_rate": 9e-06,
|
| 223 |
+
"loss": 0.1567,
|
| 224 |
+
"step": 3600
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.16,
|
| 228 |
+
"learning_rate": 9.25e-06,
|
| 229 |
+
"loss": 0.1552,
|
| 230 |
+
"step": 3700
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.19,
|
| 234 |
+
"learning_rate": 9.5e-06,
|
| 235 |
+
"loss": 0.1567,
|
| 236 |
+
"step": 3800
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.22,
|
| 240 |
+
"learning_rate": 9.750000000000002e-06,
|
| 241 |
+
"loss": 0.1557,
|
| 242 |
+
"step": 3900
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.25,
|
| 246 |
+
"learning_rate": 1e-05,
|
| 247 |
+
"loss": 0.1539,
|
| 248 |
+
"step": 4000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.28,
|
| 252 |
+
"learning_rate": 1.025e-05,
|
| 253 |
+
"loss": 0.1541,
|
| 254 |
+
"step": 4100
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.32,
|
| 258 |
+
"learning_rate": 1.05e-05,
|
| 259 |
+
"loss": 0.1518,
|
| 260 |
+
"step": 4200
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 1.35,
|
| 264 |
+
"learning_rate": 1.075e-05,
|
| 265 |
+
"loss": 0.1513,
|
| 266 |
+
"step": 4300
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.38,
|
| 270 |
+
"learning_rate": 1.1000000000000001e-05,
|
| 271 |
+
"loss": 0.1503,
|
| 272 |
+
"step": 4400
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 1.41,
|
| 276 |
+
"learning_rate": 1.125e-05,
|
| 277 |
+
"loss": 0.1506,
|
| 278 |
+
"step": 4500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.44,
|
| 282 |
+
"learning_rate": 1.1500000000000002e-05,
|
| 283 |
+
"loss": 0.1519,
|
| 284 |
+
"step": 4600
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.47,
|
| 288 |
+
"learning_rate": 1.175e-05,
|
| 289 |
+
"loss": 0.152,
|
| 290 |
+
"step": 4700
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.5,
|
| 294 |
+
"learning_rate": 1.2e-05,
|
| 295 |
+
"loss": 0.1488,
|
| 296 |
+
"step": 4800
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.54,
|
| 300 |
+
"learning_rate": 1.225e-05,
|
| 301 |
+
"loss": 0.1471,
|
| 302 |
+
"step": 4900
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.57,
|
| 306 |
+
"learning_rate": 1.25e-05,
|
| 307 |
+
"loss": 0.1476,
|
| 308 |
+
"step": 5000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.57,
|
| 312 |
+
"eval_loss": 0.2646298408508301,
|
| 313 |
+
"eval_runtime": 97.2387,
|
| 314 |
+
"eval_samples_per_second": 91.949,
|
| 315 |
+
"eval_steps_per_second": 1.44,
|
| 316 |
+
"step": 5000
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 1.6,
|
| 320 |
+
"learning_rate": 1.2478755948334468e-05,
|
| 321 |
+
"loss": 0.1465,
|
| 322 |
+
"step": 5100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 1.63,
|
| 326 |
+
"learning_rate": 1.2457511896668933e-05,
|
| 327 |
+
"loss": 0.1463,
|
| 328 |
+
"step": 5200
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 1.66,
|
| 332 |
+
"learning_rate": 1.24362678450034e-05,
|
| 333 |
+
"loss": 0.1429,
|
| 334 |
+
"step": 5300
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 1.69,
|
| 338 |
+
"learning_rate": 1.2415023793337865e-05,
|
| 339 |
+
"loss": 0.1439,
|
| 340 |
+
"step": 5400
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 1.72,
|
| 344 |
+
"learning_rate": 1.2393779741672332e-05,
|
| 345 |
+
"loss": 0.1429,
|
| 346 |
+
"step": 5500
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.75,
|
| 350 |
+
"learning_rate": 1.23725356900068e-05,
|
| 351 |
+
"loss": 0.14,
|
| 352 |
+
"step": 5600
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 1.79,
|
| 356 |
+
"learning_rate": 1.2351291638341265e-05,
|
| 357 |
+
"loss": 0.1395,
|
| 358 |
+
"step": 5700
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 1.82,
|
| 362 |
+
"learning_rate": 1.2330047586675732e-05,
|
| 363 |
+
"loss": 0.139,
|
| 364 |
+
"step": 5800
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 1.85,
|
| 368 |
+
"learning_rate": 1.2308803535010199e-05,
|
| 369 |
+
"loss": 0.137,
|
| 370 |
+
"step": 5900
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 1.88,
|
| 374 |
+
"learning_rate": 1.2287559483344664e-05,
|
| 375 |
+
"loss": 0.1336,
|
| 376 |
+
"step": 6000
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 1.91,
|
| 380 |
+
"learning_rate": 1.2266315431679131e-05,
|
| 381 |
+
"loss": 0.1368,
|
| 382 |
+
"step": 6100
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 1.94,
|
| 386 |
+
"learning_rate": 1.2245071380013596e-05,
|
| 387 |
+
"loss": 0.1348,
|
| 388 |
+
"step": 6200
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.97,
|
| 392 |
+
"learning_rate": 1.2223827328348063e-05,
|
| 393 |
+
"loss": 0.1337,
|
| 394 |
+
"step": 6300
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 2.01,
|
| 398 |
+
"learning_rate": 1.2202583276682529e-05,
|
| 399 |
+
"loss": 0.1338,
|
| 400 |
+
"step": 6400
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 2.04,
|
| 404 |
+
"learning_rate": 1.2181339225016996e-05,
|
| 405 |
+
"loss": 0.1259,
|
| 406 |
+
"step": 6500
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 2.07,
|
| 410 |
+
"learning_rate": 1.2160095173351463e-05,
|
| 411 |
+
"loss": 0.124,
|
| 412 |
+
"step": 6600
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 2.1,
|
| 416 |
+
"learning_rate": 1.213885112168593e-05,
|
| 417 |
+
"loss": 0.1244,
|
| 418 |
+
"step": 6700
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 2.13,
|
| 422 |
+
"learning_rate": 1.2117607070020395e-05,
|
| 423 |
+
"loss": 0.123,
|
| 424 |
+
"step": 6800
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 2.16,
|
| 428 |
+
"learning_rate": 1.2096363018354862e-05,
|
| 429 |
+
"loss": 0.1206,
|
| 430 |
+
"step": 6900
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 2.19,
|
| 434 |
+
"learning_rate": 1.2075118966689327e-05,
|
| 435 |
+
"loss": 0.1229,
|
| 436 |
+
"step": 7000
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 2.22,
|
| 440 |
+
"learning_rate": 1.2053874915023794e-05,
|
| 441 |
+
"loss": 0.1204,
|
| 442 |
+
"step": 7100
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 2.26,
|
| 446 |
+
"learning_rate": 1.203263086335826e-05,
|
| 447 |
+
"loss": 0.1223,
|
| 448 |
+
"step": 7200
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 2.29,
|
| 452 |
+
"learning_rate": 1.2011386811692727e-05,
|
| 453 |
+
"loss": 0.12,
|
| 454 |
+
"step": 7300
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 2.32,
|
| 458 |
+
"learning_rate": 1.1990142760027192e-05,
|
| 459 |
+
"loss": 0.1193,
|
| 460 |
+
"step": 7400
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 2.35,
|
| 464 |
+
"learning_rate": 1.196889870836166e-05,
|
| 465 |
+
"loss": 0.117,
|
| 466 |
+
"step": 7500
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 2.38,
|
| 470 |
+
"learning_rate": 1.1947654656696125e-05,
|
| 471 |
+
"loss": 0.1184,
|
| 472 |
+
"step": 7600
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 2.41,
|
| 476 |
+
"learning_rate": 1.1926410605030593e-05,
|
| 477 |
+
"loss": 0.1187,
|
| 478 |
+
"step": 7700
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 2.44,
|
| 482 |
+
"learning_rate": 1.1905166553365059e-05,
|
| 483 |
+
"loss": 0.1171,
|
| 484 |
+
"step": 7800
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 2.48,
|
| 488 |
+
"learning_rate": 1.1883922501699526e-05,
|
| 489 |
+
"loss": 0.1156,
|
| 490 |
+
"step": 7900
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 2.51,
|
| 494 |
+
"learning_rate": 1.1862678450033991e-05,
|
| 495 |
+
"loss": 0.1171,
|
| 496 |
+
"step": 8000
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 2.54,
|
| 500 |
+
"learning_rate": 1.1841434398368458e-05,
|
| 501 |
+
"loss": 0.1176,
|
| 502 |
+
"step": 8100
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 2.57,
|
| 506 |
+
"learning_rate": 1.1820190346702923e-05,
|
| 507 |
+
"loss": 0.1149,
|
| 508 |
+
"step": 8200
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 2.6,
|
| 512 |
+
"learning_rate": 1.179894629503739e-05,
|
| 513 |
+
"loss": 0.1129,
|
| 514 |
+
"step": 8300
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 2.63,
|
| 518 |
+
"learning_rate": 1.1777702243371856e-05,
|
| 519 |
+
"loss": 0.1162,
|
| 520 |
+
"step": 8400
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 2.66,
|
| 524 |
+
"learning_rate": 1.1756458191706323e-05,
|
| 525 |
+
"loss": 0.1127,
|
| 526 |
+
"step": 8500
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 2.69,
|
| 530 |
+
"learning_rate": 1.1735214140040788e-05,
|
| 531 |
+
"loss": 0.1146,
|
| 532 |
+
"step": 8600
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 2.73,
|
| 536 |
+
"learning_rate": 1.1713970088375255e-05,
|
| 537 |
+
"loss": 0.1112,
|
| 538 |
+
"step": 8700
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 2.76,
|
| 542 |
+
"learning_rate": 1.1692726036709722e-05,
|
| 543 |
+
"loss": 0.1123,
|
| 544 |
+
"step": 8800
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 2.79,
|
| 548 |
+
"learning_rate": 1.1671481985044189e-05,
|
| 549 |
+
"loss": 0.1127,
|
| 550 |
+
"step": 8900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 2.82,
|
| 554 |
+
"learning_rate": 1.1650237933378654e-05,
|
| 555 |
+
"loss": 0.1132,
|
| 556 |
+
"step": 9000
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 2.85,
|
| 560 |
+
"learning_rate": 1.1628993881713121e-05,
|
| 561 |
+
"loss": 0.1118,
|
| 562 |
+
"step": 9100
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 2.88,
|
| 566 |
+
"learning_rate": 1.1607749830047587e-05,
|
| 567 |
+
"loss": 0.1121,
|
| 568 |
+
"step": 9200
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 2.91,
|
| 572 |
+
"learning_rate": 1.1586505778382054e-05,
|
| 573 |
+
"loss": 0.1103,
|
| 574 |
+
"step": 9300
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 2.95,
|
| 578 |
+
"learning_rate": 1.1565261726716519e-05,
|
| 579 |
+
"loss": 0.1098,
|
| 580 |
+
"step": 9400
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 2.98,
|
| 584 |
+
"learning_rate": 1.1544017675050986e-05,
|
| 585 |
+
"loss": 0.1071,
|
| 586 |
+
"step": 9500
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 3.01,
|
| 590 |
+
"learning_rate": 1.1522773623385451e-05,
|
| 591 |
+
"loss": 0.1065,
|
| 592 |
+
"step": 9600
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 3.04,
|
| 596 |
+
"learning_rate": 1.1501529571719918e-05,
|
| 597 |
+
"loss": 0.1019,
|
| 598 |
+
"step": 9700
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 3.07,
|
| 602 |
+
"learning_rate": 1.1480285520054385e-05,
|
| 603 |
+
"loss": 0.1017,
|
| 604 |
+
"step": 9800
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 3.1,
|
| 608 |
+
"learning_rate": 1.1459041468388852e-05,
|
| 609 |
+
"loss": 0.1023,
|
| 610 |
+
"step": 9900
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 3.13,
|
| 614 |
+
"learning_rate": 1.1437797416723318e-05,
|
| 615 |
+
"loss": 0.0995,
|
| 616 |
+
"step": 10000
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 3.13,
|
| 620 |
+
"eval_loss": 0.21612465381622314,
|
| 621 |
+
"eval_runtime": 83.5806,
|
| 622 |
+
"eval_samples_per_second": 106.975,
|
| 623 |
+
"eval_steps_per_second": 1.675,
|
| 624 |
+
"step": 10000
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 3.16,
|
| 628 |
+
"learning_rate": 1.1416553365057785e-05,
|
| 629 |
+
"loss": 0.1016,
|
| 630 |
+
"step": 10100
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 3.2,
|
| 634 |
+
"learning_rate": 1.139530931339225e-05,
|
| 635 |
+
"loss": 0.0998,
|
| 636 |
+
"step": 10200
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 3.23,
|
| 640 |
+
"learning_rate": 1.1374065261726717e-05,
|
| 641 |
+
"loss": 0.1009,
|
| 642 |
+
"step": 10300
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 3.26,
|
| 646 |
+
"learning_rate": 1.1352821210061182e-05,
|
| 647 |
+
"loss": 0.0986,
|
| 648 |
+
"step": 10400
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 3.29,
|
| 652 |
+
"learning_rate": 1.133157715839565e-05,
|
| 653 |
+
"loss": 0.1003,
|
| 654 |
+
"step": 10500
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 3.32,
|
| 658 |
+
"learning_rate": 1.1310333106730117e-05,
|
| 659 |
+
"loss": 0.1005,
|
| 660 |
+
"step": 10600
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 3.35,
|
| 664 |
+
"learning_rate": 1.1289089055064582e-05,
|
| 665 |
+
"loss": 0.1012,
|
| 666 |
+
"step": 10700
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 3.38,
|
| 670 |
+
"learning_rate": 1.1267845003399049e-05,
|
| 671 |
+
"loss": 0.1005,
|
| 672 |
+
"step": 10800
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 3.42,
|
| 676 |
+
"learning_rate": 1.1246600951733516e-05,
|
| 677 |
+
"loss": 0.1,
|
| 678 |
+
"step": 10900
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 3.45,
|
| 682 |
+
"learning_rate": 1.1225356900067981e-05,
|
| 683 |
+
"loss": 0.0994,
|
| 684 |
+
"step": 11000
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 3.48,
|
| 688 |
+
"learning_rate": 1.1204112848402448e-05,
|
| 689 |
+
"loss": 0.0972,
|
| 690 |
+
"step": 11100
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 3.51,
|
| 694 |
+
"learning_rate": 1.1182868796736915e-05,
|
| 695 |
+
"loss": 0.0981,
|
| 696 |
+
"step": 11200
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 3.54,
|
| 700 |
+
"learning_rate": 1.116162474507138e-05,
|
| 701 |
+
"loss": 0.0956,
|
| 702 |
+
"step": 11300
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 3.57,
|
| 706 |
+
"learning_rate": 1.1140380693405848e-05,
|
| 707 |
+
"loss": 0.0955,
|
| 708 |
+
"step": 11400
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 3.6,
|
| 712 |
+
"learning_rate": 1.1119136641740313e-05,
|
| 713 |
+
"loss": 0.0955,
|
| 714 |
+
"step": 11500
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 3.63,
|
| 718 |
+
"learning_rate": 1.109789259007478e-05,
|
| 719 |
+
"loss": 0.0978,
|
| 720 |
+
"step": 11600
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 3.67,
|
| 724 |
+
"learning_rate": 1.1076648538409245e-05,
|
| 725 |
+
"loss": 0.0959,
|
| 726 |
+
"step": 11700
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 3.7,
|
| 730 |
+
"learning_rate": 1.1055404486743712e-05,
|
| 731 |
+
"loss": 0.0968,
|
| 732 |
+
"step": 11800
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 3.73,
|
| 736 |
+
"learning_rate": 1.1034160435078178e-05,
|
| 737 |
+
"loss": 0.096,
|
| 738 |
+
"step": 11900
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 3.76,
|
| 742 |
+
"learning_rate": 1.1012916383412646e-05,
|
| 743 |
+
"loss": 0.0959,
|
| 744 |
+
"step": 12000
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 3.79,
|
| 748 |
+
"learning_rate": 1.0991672331747112e-05,
|
| 749 |
+
"loss": 0.0959,
|
| 750 |
+
"step": 12100
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 3.82,
|
| 754 |
+
"learning_rate": 1.0970428280081579e-05,
|
| 755 |
+
"loss": 0.0959,
|
| 756 |
+
"step": 12200
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 3.85,
|
| 760 |
+
"learning_rate": 1.0949184228416044e-05,
|
| 761 |
+
"loss": 0.0957,
|
| 762 |
+
"step": 12300
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 3.89,
|
| 766 |
+
"learning_rate": 1.0927940176750511e-05,
|
| 767 |
+
"loss": 0.0948,
|
| 768 |
+
"step": 12400
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 3.92,
|
| 772 |
+
"learning_rate": 1.0906696125084976e-05,
|
| 773 |
+
"loss": 0.0959,
|
| 774 |
+
"step": 12500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 3.95,
|
| 778 |
+
"learning_rate": 1.0885452073419443e-05,
|
| 779 |
+
"loss": 0.0942,
|
| 780 |
+
"step": 12600
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 3.98,
|
| 784 |
+
"learning_rate": 1.0864208021753909e-05,
|
| 785 |
+
"loss": 0.0941,
|
| 786 |
+
"step": 12700
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 4.01,
|
| 790 |
+
"learning_rate": 1.0842963970088376e-05,
|
| 791 |
+
"loss": 0.0905,
|
| 792 |
+
"step": 12800
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 4.04,
|
| 796 |
+
"learning_rate": 1.0821719918422841e-05,
|
| 797 |
+
"loss": 0.0887,
|
| 798 |
+
"step": 12900
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 4.07,
|
| 802 |
+
"learning_rate": 1.080047586675731e-05,
|
| 803 |
+
"loss": 0.0873,
|
| 804 |
+
"step": 13000
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 4.1,
|
| 808 |
+
"learning_rate": 1.0779231815091775e-05,
|
| 809 |
+
"loss": 0.0887,
|
| 810 |
+
"step": 13100
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 4.14,
|
| 814 |
+
"learning_rate": 1.0757987763426242e-05,
|
| 815 |
+
"loss": 0.0879,
|
| 816 |
+
"step": 13200
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 4.17,
|
| 820 |
+
"learning_rate": 1.0736743711760707e-05,
|
| 821 |
+
"loss": 0.0889,
|
| 822 |
+
"step": 13300
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 4.2,
|
| 826 |
+
"learning_rate": 1.0715499660095175e-05,
|
| 827 |
+
"loss": 0.0869,
|
| 828 |
+
"step": 13400
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 4.23,
|
| 832 |
+
"learning_rate": 1.069425560842964e-05,
|
| 833 |
+
"loss": 0.0865,
|
| 834 |
+
"step": 13500
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 4.26,
|
| 838 |
+
"learning_rate": 1.0673011556764107e-05,
|
| 839 |
+
"loss": 0.0863,
|
| 840 |
+
"step": 13600
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 4.29,
|
| 844 |
+
"learning_rate": 1.0651767505098572e-05,
|
| 845 |
+
"loss": 0.0873,
|
| 846 |
+
"step": 13700
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 4.32,
|
| 850 |
+
"learning_rate": 1.063052345343304e-05,
|
| 851 |
+
"loss": 0.0856,
|
| 852 |
+
"step": 13800
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 4.36,
|
| 856 |
+
"learning_rate": 1.0609279401767505e-05,
|
| 857 |
+
"loss": 0.0872,
|
| 858 |
+
"step": 13900
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 4.39,
|
| 862 |
+
"learning_rate": 1.0588035350101972e-05,
|
| 863 |
+
"loss": 0.0865,
|
| 864 |
+
"step": 14000
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 4.42,
|
| 868 |
+
"learning_rate": 1.0566791298436439e-05,
|
| 869 |
+
"loss": 0.0868,
|
| 870 |
+
"step": 14100
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 4.45,
|
| 874 |
+
"learning_rate": 1.0545547246770906e-05,
|
| 875 |
+
"loss": 0.0858,
|
| 876 |
+
"step": 14200
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 4.48,
|
| 880 |
+
"learning_rate": 1.0524303195105371e-05,
|
| 881 |
+
"loss": 0.0863,
|
| 882 |
+
"step": 14300
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 4.51,
|
| 886 |
+
"learning_rate": 1.0503059143439838e-05,
|
| 887 |
+
"loss": 0.0861,
|
| 888 |
+
"step": 14400
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 4.54,
|
| 892 |
+
"learning_rate": 1.0481815091774303e-05,
|
| 893 |
+
"loss": 0.0855,
|
| 894 |
+
"step": 14500
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 4.57,
|
| 898 |
+
"learning_rate": 1.046057104010877e-05,
|
| 899 |
+
"loss": 0.0874,
|
| 900 |
+
"step": 14600
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 4.61,
|
| 904 |
+
"learning_rate": 1.0439326988443236e-05,
|
| 905 |
+
"loss": 0.0858,
|
| 906 |
+
"step": 14700
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 4.64,
|
| 910 |
+
"learning_rate": 1.0418082936777703e-05,
|
| 911 |
+
"loss": 0.0848,
|
| 912 |
+
"step": 14800
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 4.67,
|
| 916 |
+
"learning_rate": 1.0396838885112168e-05,
|
| 917 |
+
"loss": 0.0846,
|
| 918 |
+
"step": 14900
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 4.7,
|
| 922 |
+
"learning_rate": 1.0375594833446635e-05,
|
| 923 |
+
"loss": 0.0835,
|
| 924 |
+
"step": 15000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 4.7,
|
| 928 |
+
"eval_loss": 0.1958872377872467,
|
| 929 |
+
"eval_runtime": 88.1156,
|
| 930 |
+
"eval_samples_per_second": 101.469,
|
| 931 |
+
"eval_steps_per_second": 1.589,
|
| 932 |
+
"step": 15000
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 4.73,
|
| 936 |
+
"learning_rate": 1.03543507817811e-05,
|
| 937 |
+
"loss": 0.0854,
|
| 938 |
+
"step": 15100
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 4.76,
|
| 942 |
+
"learning_rate": 1.0333106730115569e-05,
|
| 943 |
+
"loss": 0.0835,
|
| 944 |
+
"step": 15200
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 4.79,
|
| 948 |
+
"learning_rate": 1.0311862678450034e-05,
|
| 949 |
+
"loss": 0.0849,
|
| 950 |
+
"step": 15300
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 4.83,
|
| 954 |
+
"learning_rate": 1.0290618626784501e-05,
|
| 955 |
+
"loss": 0.0825,
|
| 956 |
+
"step": 15400
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 4.86,
|
| 960 |
+
"learning_rate": 1.0269374575118967e-05,
|
| 961 |
+
"loss": 0.0859,
|
| 962 |
+
"step": 15500
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 4.89,
|
| 966 |
+
"learning_rate": 1.0248130523453434e-05,
|
| 967 |
+
"loss": 0.0837,
|
| 968 |
+
"step": 15600
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 4.92,
|
| 972 |
+
"learning_rate": 1.0226886471787899e-05,
|
| 973 |
+
"loss": 0.0843,
|
| 974 |
+
"step": 15700
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 4.95,
|
| 978 |
+
"learning_rate": 1.0205642420122366e-05,
|
| 979 |
+
"loss": 0.0826,
|
| 980 |
+
"step": 15800
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 4.98,
|
| 984 |
+
"learning_rate": 1.0184398368456833e-05,
|
| 985 |
+
"loss": 0.0838,
|
| 986 |
+
"step": 15900
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 5.01,
|
| 990 |
+
"learning_rate": 1.0163154316791298e-05,
|
| 991 |
+
"loss": 0.081,
|
| 992 |
+
"step": 16000
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 5.04,
|
| 996 |
+
"learning_rate": 1.0141910265125765e-05,
|
| 997 |
+
"loss": 0.0767,
|
| 998 |
+
"step": 16100
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 5.08,
|
| 1002 |
+
"learning_rate": 1.0120666213460232e-05,
|
| 1003 |
+
"loss": 0.0778,
|
| 1004 |
+
"step": 16200
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 5.11,
|
| 1008 |
+
"learning_rate": 1.0099422161794698e-05,
|
| 1009 |
+
"loss": 0.0767,
|
| 1010 |
+
"step": 16300
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 5.14,
|
| 1014 |
+
"learning_rate": 1.0078178110129165e-05,
|
| 1015 |
+
"loss": 0.0775,
|
| 1016 |
+
"step": 16400
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 5.17,
|
| 1020 |
+
"learning_rate": 1.005693405846363e-05,
|
| 1021 |
+
"loss": 0.0774,
|
| 1022 |
+
"step": 16500
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 5.2,
|
| 1026 |
+
"learning_rate": 1.0035690006798097e-05,
|
| 1027 |
+
"loss": 0.0775,
|
| 1028 |
+
"step": 16600
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 5.23,
|
| 1032 |
+
"learning_rate": 1.0014445955132564e-05,
|
| 1033 |
+
"loss": 0.0778,
|
| 1034 |
+
"step": 16700
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 5.26,
|
| 1038 |
+
"learning_rate": 9.99320190346703e-06,
|
| 1039 |
+
"loss": 0.078,
|
| 1040 |
+
"step": 16800
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 5.3,
|
| 1044 |
+
"learning_rate": 9.971957851801497e-06,
|
| 1045 |
+
"loss": 0.0777,
|
| 1046 |
+
"step": 16900
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 5.33,
|
| 1050 |
+
"learning_rate": 9.950713800135962e-06,
|
| 1051 |
+
"loss": 0.0793,
|
| 1052 |
+
"step": 17000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 5.36,
|
| 1056 |
+
"learning_rate": 9.929469748470429e-06,
|
| 1057 |
+
"loss": 0.0765,
|
| 1058 |
+
"step": 17100
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 5.39,
|
| 1062 |
+
"learning_rate": 9.908225696804894e-06,
|
| 1063 |
+
"loss": 0.0771,
|
| 1064 |
+
"step": 17200
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 5.42,
|
| 1068 |
+
"learning_rate": 9.886981645139363e-06,
|
| 1069 |
+
"loss": 0.0768,
|
| 1070 |
+
"step": 17300
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 5.45,
|
| 1074 |
+
"learning_rate": 9.865737593473828e-06,
|
| 1075 |
+
"loss": 0.0761,
|
| 1076 |
+
"step": 17400
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 5.48,
|
| 1080 |
+
"learning_rate": 9.844493541808295e-06,
|
| 1081 |
+
"loss": 0.0778,
|
| 1082 |
+
"step": 17500
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 5.51,
|
| 1086 |
+
"learning_rate": 9.82324949014276e-06,
|
| 1087 |
+
"loss": 0.0787,
|
| 1088 |
+
"step": 17600
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 5.55,
|
| 1092 |
+
"learning_rate": 9.802005438477228e-06,
|
| 1093 |
+
"loss": 0.0752,
|
| 1094 |
+
"step": 17700
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 5.58,
|
| 1098 |
+
"learning_rate": 9.780761386811693e-06,
|
| 1099 |
+
"loss": 0.0756,
|
| 1100 |
+
"step": 17800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 5.61,
|
| 1104 |
+
"learning_rate": 9.75951733514616e-06,
|
| 1105 |
+
"loss": 0.0761,
|
| 1106 |
+
"step": 17900
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 5.64,
|
| 1110 |
+
"learning_rate": 9.738273283480625e-06,
|
| 1111 |
+
"loss": 0.0776,
|
| 1112 |
+
"step": 18000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 5.67,
|
| 1116 |
+
"learning_rate": 9.717029231815092e-06,
|
| 1117 |
+
"loss": 0.0746,
|
| 1118 |
+
"step": 18100
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 5.7,
|
| 1122 |
+
"learning_rate": 9.695785180149558e-06,
|
| 1123 |
+
"loss": 0.0765,
|
| 1124 |
+
"step": 18200
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 5.73,
|
| 1128 |
+
"learning_rate": 9.674541128484025e-06,
|
| 1129 |
+
"loss": 0.0762,
|
| 1130 |
+
"step": 18300
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 5.77,
|
| 1134 |
+
"learning_rate": 9.653297076818492e-06,
|
| 1135 |
+
"loss": 0.0763,
|
| 1136 |
+
"step": 18400
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 5.8,
|
| 1140 |
+
"learning_rate": 9.632053025152959e-06,
|
| 1141 |
+
"loss": 0.0754,
|
| 1142 |
+
"step": 18500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 5.83,
|
| 1146 |
+
"learning_rate": 9.610808973487424e-06,
|
| 1147 |
+
"loss": 0.0772,
|
| 1148 |
+
"step": 18600
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 5.86,
|
| 1152 |
+
"learning_rate": 9.589564921821891e-06,
|
| 1153 |
+
"loss": 0.0739,
|
| 1154 |
+
"step": 18700
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 5.89,
|
| 1158 |
+
"learning_rate": 9.568320870156356e-06,
|
| 1159 |
+
"loss": 0.0732,
|
| 1160 |
+
"step": 18800
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 5.92,
|
| 1164 |
+
"learning_rate": 9.547076818490823e-06,
|
| 1165 |
+
"loss": 0.0756,
|
| 1166 |
+
"step": 18900
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 5.95,
|
| 1170 |
+
"learning_rate": 9.525832766825289e-06,
|
| 1171 |
+
"loss": 0.0757,
|
| 1172 |
+
"step": 19000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 5.98,
|
| 1176 |
+
"learning_rate": 9.504588715159756e-06,
|
| 1177 |
+
"loss": 0.0747,
|
| 1178 |
+
"step": 19100
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 6.02,
|
| 1182 |
+
"learning_rate": 9.483344663494221e-06,
|
| 1183 |
+
"loss": 0.0721,
|
| 1184 |
+
"step": 19200
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 6.05,
|
| 1188 |
+
"learning_rate": 9.462100611828688e-06,
|
| 1189 |
+
"loss": 0.0695,
|
| 1190 |
+
"step": 19300
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 6.08,
|
| 1194 |
+
"learning_rate": 9.440856560163155e-06,
|
| 1195 |
+
"loss": 0.0694,
|
| 1196 |
+
"step": 19400
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 6.11,
|
| 1200 |
+
"learning_rate": 9.419612508497622e-06,
|
| 1201 |
+
"loss": 0.0696,
|
| 1202 |
+
"step": 19500
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 6.14,
|
| 1206 |
+
"learning_rate": 9.398368456832087e-06,
|
| 1207 |
+
"loss": 0.0695,
|
| 1208 |
+
"step": 19600
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 6.17,
|
| 1212 |
+
"learning_rate": 9.377124405166555e-06,
|
| 1213 |
+
"loss": 0.069,
|
| 1214 |
+
"step": 19700
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 6.2,
|
| 1218 |
+
"learning_rate": 9.35588035350102e-06,
|
| 1219 |
+
"loss": 0.0708,
|
| 1220 |
+
"step": 19800
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 6.24,
|
| 1224 |
+
"learning_rate": 9.334636301835487e-06,
|
| 1225 |
+
"loss": 0.07,
|
| 1226 |
+
"step": 19900
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 6.27,
|
| 1230 |
+
"learning_rate": 9.313392250169952e-06,
|
| 1231 |
+
"loss": 0.0706,
|
| 1232 |
+
"step": 20000
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 6.27,
|
| 1236 |
+
"eval_loss": 0.18451732397079468,
|
| 1237 |
+
"eval_runtime": 91.4249,
|
| 1238 |
+
"eval_samples_per_second": 97.796,
|
| 1239 |
+
"eval_steps_per_second": 1.531,
|
| 1240 |
+
"step": 20000
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 6.3,
|
| 1244 |
+
"learning_rate": 9.29214819850442e-06,
|
| 1245 |
+
"loss": 0.0692,
|
| 1246 |
+
"step": 20100
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 6.33,
|
| 1250 |
+
"learning_rate": 9.270904146838885e-06,
|
| 1251 |
+
"loss": 0.0681,
|
| 1252 |
+
"step": 20200
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 6.36,
|
| 1256 |
+
"learning_rate": 9.249660095173352e-06,
|
| 1257 |
+
"loss": 0.0712,
|
| 1258 |
+
"step": 20300
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 6.39,
|
| 1262 |
+
"learning_rate": 9.228416043507817e-06,
|
| 1263 |
+
"loss": 0.0683,
|
| 1264 |
+
"step": 20400
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 6.42,
|
| 1268 |
+
"learning_rate": 9.207171991842286e-06,
|
| 1269 |
+
"loss": 0.0684,
|
| 1270 |
+
"step": 20500
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 6.45,
|
| 1274 |
+
"learning_rate": 9.185927940176751e-06,
|
| 1275 |
+
"loss": 0.0688,
|
| 1276 |
+
"step": 20600
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 6.49,
|
| 1280 |
+
"learning_rate": 9.164683888511218e-06,
|
| 1281 |
+
"loss": 0.0687,
|
| 1282 |
+
"step": 20700
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 6.52,
|
| 1286 |
+
"learning_rate": 9.143439836845683e-06,
|
| 1287 |
+
"loss": 0.0695,
|
| 1288 |
+
"step": 20800
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 6.55,
|
| 1292 |
+
"learning_rate": 9.12219578518015e-06,
|
| 1293 |
+
"loss": 0.0695,
|
| 1294 |
+
"step": 20900
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 6.58,
|
| 1298 |
+
"learning_rate": 9.100951733514616e-06,
|
| 1299 |
+
"loss": 0.069,
|
| 1300 |
+
"step": 21000
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 6.61,
|
| 1304 |
+
"learning_rate": 9.079707681849083e-06,
|
| 1305 |
+
"loss": 0.0684,
|
| 1306 |
+
"step": 21100
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 6.64,
|
| 1310 |
+
"learning_rate": 9.058463630183548e-06,
|
| 1311 |
+
"loss": 0.0675,
|
| 1312 |
+
"step": 21200
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 6.67,
|
| 1316 |
+
"learning_rate": 9.037219578518015e-06,
|
| 1317 |
+
"loss": 0.07,
|
| 1318 |
+
"step": 21300
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 6.71,
|
| 1322 |
+
"learning_rate": 9.015975526852482e-06,
|
| 1323 |
+
"loss": 0.0687,
|
| 1324 |
+
"step": 21400
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 6.74,
|
| 1328 |
+
"learning_rate": 8.994731475186947e-06,
|
| 1329 |
+
"loss": 0.0683,
|
| 1330 |
+
"step": 21500
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 6.77,
|
| 1334 |
+
"learning_rate": 8.973487423521414e-06,
|
| 1335 |
+
"loss": 0.0686,
|
| 1336 |
+
"step": 21600
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 6.8,
|
| 1340 |
+
"learning_rate": 8.952243371855881e-06,
|
| 1341 |
+
"loss": 0.0687,
|
| 1342 |
+
"step": 21700
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 6.83,
|
| 1346 |
+
"learning_rate": 8.930999320190347e-06,
|
| 1347 |
+
"loss": 0.0682,
|
| 1348 |
+
"step": 21800
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 6.86,
|
| 1352 |
+
"learning_rate": 8.909755268524814e-06,
|
| 1353 |
+
"loss": 0.0698,
|
| 1354 |
+
"step": 21900
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 6.89,
|
| 1358 |
+
"learning_rate": 8.888511216859279e-06,
|
| 1359 |
+
"loss": 0.068,
|
| 1360 |
+
"step": 22000
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 6.92,
|
| 1364 |
+
"learning_rate": 8.867267165193746e-06,
|
| 1365 |
+
"loss": 0.068,
|
| 1366 |
+
"step": 22100
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 6.96,
|
| 1370 |
+
"learning_rate": 8.846023113528213e-06,
|
| 1371 |
+
"loss": 0.0665,
|
| 1372 |
+
"step": 22200
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 6.99,
|
| 1376 |
+
"learning_rate": 8.824779061862678e-06,
|
| 1377 |
+
"loss": 0.0701,
|
| 1378 |
+
"step": 22300
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 7.02,
|
| 1382 |
+
"learning_rate": 8.803535010197145e-06,
|
| 1383 |
+
"loss": 0.0649,
|
| 1384 |
+
"step": 22400
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 7.05,
|
| 1388 |
+
"learning_rate": 8.78229095853161e-06,
|
| 1389 |
+
"loss": 0.0632,
|
| 1390 |
+
"step": 22500
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 7.08,
|
| 1394 |
+
"learning_rate": 8.761046906866078e-06,
|
| 1395 |
+
"loss": 0.0641,
|
| 1396 |
+
"step": 22600
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 7.11,
|
| 1400 |
+
"learning_rate": 8.739802855200545e-06,
|
| 1401 |
+
"loss": 0.0611,
|
| 1402 |
+
"step": 22700
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 7.14,
|
| 1406 |
+
"learning_rate": 8.718558803535012e-06,
|
| 1407 |
+
"loss": 0.0631,
|
| 1408 |
+
"step": 22800
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 7.18,
|
| 1412 |
+
"learning_rate": 8.697314751869477e-06,
|
| 1413 |
+
"loss": 0.0644,
|
| 1414 |
+
"step": 22900
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 7.21,
|
| 1418 |
+
"learning_rate": 8.676070700203944e-06,
|
| 1419 |
+
"loss": 0.0607,
|
| 1420 |
+
"step": 23000
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 7.24,
|
| 1424 |
+
"learning_rate": 8.65482664853841e-06,
|
| 1425 |
+
"loss": 0.0627,
|
| 1426 |
+
"step": 23100
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 7.27,
|
| 1430 |
+
"learning_rate": 8.633582596872877e-06,
|
| 1431 |
+
"loss": 0.0635,
|
| 1432 |
+
"step": 23200
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 7.3,
|
| 1436 |
+
"learning_rate": 8.612338545207342e-06,
|
| 1437 |
+
"loss": 0.0635,
|
| 1438 |
+
"step": 23300
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 7.33,
|
| 1442 |
+
"learning_rate": 8.591094493541809e-06,
|
| 1443 |
+
"loss": 0.0618,
|
| 1444 |
+
"step": 23400
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 7.36,
|
| 1448 |
+
"learning_rate": 8.569850441876274e-06,
|
| 1449 |
+
"loss": 0.0625,
|
| 1450 |
+
"step": 23500
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 7.39,
|
| 1454 |
+
"learning_rate": 8.548606390210741e-06,
|
| 1455 |
+
"loss": 0.0646,
|
| 1456 |
+
"step": 23600
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 7.43,
|
| 1460 |
+
"learning_rate": 8.527362338545208e-06,
|
| 1461 |
+
"loss": 0.0628,
|
| 1462 |
+
"step": 23700
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 7.46,
|
| 1466 |
+
"learning_rate": 8.506118286879675e-06,
|
| 1467 |
+
"loss": 0.063,
|
| 1468 |
+
"step": 23800
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 7.49,
|
| 1472 |
+
"learning_rate": 8.48487423521414e-06,
|
| 1473 |
+
"loss": 0.0625,
|
| 1474 |
+
"step": 23900
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 7.52,
|
| 1478 |
+
"learning_rate": 8.463630183548608e-06,
|
| 1479 |
+
"loss": 0.0632,
|
| 1480 |
+
"step": 24000
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 7.55,
|
| 1484 |
+
"learning_rate": 8.442386131883073e-06,
|
| 1485 |
+
"loss": 0.0634,
|
| 1486 |
+
"step": 24100
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 7.58,
|
| 1490 |
+
"learning_rate": 8.42114208021754e-06,
|
| 1491 |
+
"loss": 0.0622,
|
| 1492 |
+
"step": 24200
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 7.61,
|
| 1496 |
+
"learning_rate": 8.399898028552005e-06,
|
| 1497 |
+
"loss": 0.0625,
|
| 1498 |
+
"step": 24300
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 7.65,
|
| 1502 |
+
"learning_rate": 8.378653976886472e-06,
|
| 1503 |
+
"loss": 0.0632,
|
| 1504 |
+
"step": 24400
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 7.68,
|
| 1508 |
+
"learning_rate": 8.357409925220938e-06,
|
| 1509 |
+
"loss": 0.0627,
|
| 1510 |
+
"step": 24500
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 7.71,
|
| 1514 |
+
"learning_rate": 8.336165873555405e-06,
|
| 1515 |
+
"loss": 0.0631,
|
| 1516 |
+
"step": 24600
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 7.74,
|
| 1520 |
+
"learning_rate": 8.31492182188987e-06,
|
| 1521 |
+
"loss": 0.0633,
|
| 1522 |
+
"step": 24700
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 7.77,
|
| 1526 |
+
"learning_rate": 8.293677770224339e-06,
|
| 1527 |
+
"loss": 0.0638,
|
| 1528 |
+
"step": 24800
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 7.8,
|
| 1532 |
+
"learning_rate": 8.272433718558804e-06,
|
| 1533 |
+
"loss": 0.0626,
|
| 1534 |
+
"step": 24900
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 7.83,
|
| 1538 |
+
"learning_rate": 8.251189666893271e-06,
|
| 1539 |
+
"loss": 0.063,
|
| 1540 |
+
"step": 25000
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 7.83,
|
| 1544 |
+
"eval_loss": 0.17496153712272644,
|
| 1545 |
+
"eval_runtime": 98.6326,
|
| 1546 |
+
"eval_samples_per_second": 90.65,
|
| 1547 |
+
"eval_steps_per_second": 1.419,
|
| 1548 |
+
"step": 25000
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 7.86,
|
| 1552 |
+
"learning_rate": 8.229945615227736e-06,
|
| 1553 |
+
"loss": 0.0634,
|
| 1554 |
+
"step": 25100
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 7.9,
|
| 1558 |
+
"learning_rate": 8.208701563562203e-06,
|
| 1559 |
+
"loss": 0.063,
|
| 1560 |
+
"step": 25200
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 7.93,
|
| 1564 |
+
"learning_rate": 8.187457511896669e-06,
|
| 1565 |
+
"loss": 0.0634,
|
| 1566 |
+
"step": 25300
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 7.96,
|
| 1570 |
+
"learning_rate": 8.166213460231136e-06,
|
| 1571 |
+
"loss": 0.0636,
|
| 1572 |
+
"step": 25400
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 7.99,
|
| 1576 |
+
"learning_rate": 8.144969408565601e-06,
|
| 1577 |
+
"loss": 0.0607,
|
| 1578 |
+
"step": 25500
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 8.02,
|
| 1582 |
+
"learning_rate": 8.123725356900068e-06,
|
| 1583 |
+
"loss": 0.0576,
|
| 1584 |
+
"step": 25600
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 8.05,
|
| 1588 |
+
"learning_rate": 8.102481305234533e-06,
|
| 1589 |
+
"loss": 0.0579,
|
| 1590 |
+
"step": 25700
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 8.08,
|
| 1594 |
+
"learning_rate": 8.081237253569002e-06,
|
| 1595 |
+
"loss": 0.0581,
|
| 1596 |
+
"step": 25800
|
| 1597 |
+
},
|
| 1598 |
+
{
|
| 1599 |
+
"epoch": 8.12,
|
| 1600 |
+
"learning_rate": 8.059993201903467e-06,
|
| 1601 |
+
"loss": 0.0584,
|
| 1602 |
+
"step": 25900
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 8.15,
|
| 1606 |
+
"learning_rate": 8.038749150237935e-06,
|
| 1607 |
+
"loss": 0.0578,
|
| 1608 |
+
"step": 26000
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 8.18,
|
| 1612 |
+
"learning_rate": 8.0175050985724e-06,
|
| 1613 |
+
"loss": 0.0564,
|
| 1614 |
+
"step": 26100
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 8.21,
|
| 1618 |
+
"learning_rate": 7.996261046906867e-06,
|
| 1619 |
+
"loss": 0.0583,
|
| 1620 |
+
"step": 26200
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 8.24,
|
| 1624 |
+
"learning_rate": 7.975016995241332e-06,
|
| 1625 |
+
"loss": 0.0568,
|
| 1626 |
+
"step": 26300
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 8.27,
|
| 1630 |
+
"learning_rate": 7.9537729435758e-06,
|
| 1631 |
+
"loss": 0.0584,
|
| 1632 |
+
"step": 26400
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 8.3,
|
| 1636 |
+
"learning_rate": 7.932528891910265e-06,
|
| 1637 |
+
"loss": 0.0585,
|
| 1638 |
+
"step": 26500
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 8.33,
|
| 1642 |
+
"learning_rate": 7.911284840244732e-06,
|
| 1643 |
+
"loss": 0.0589,
|
| 1644 |
+
"step": 26600
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 8.37,
|
| 1648 |
+
"learning_rate": 7.890040788579197e-06,
|
| 1649 |
+
"loss": 0.0569,
|
| 1650 |
+
"step": 26700
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 8.4,
|
| 1654 |
+
"learning_rate": 7.868796736913664e-06,
|
| 1655 |
+
"loss": 0.0575,
|
| 1656 |
+
"step": 26800
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 8.43,
|
| 1660 |
+
"learning_rate": 7.847552685248131e-06,
|
| 1661 |
+
"loss": 0.057,
|
| 1662 |
+
"step": 26900
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 8.46,
|
| 1666 |
+
"learning_rate": 7.826308633582598e-06,
|
| 1667 |
+
"loss": 0.0591,
|
| 1668 |
+
"step": 27000
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 8.49,
|
| 1672 |
+
"learning_rate": 7.805064581917063e-06,
|
| 1673 |
+
"loss": 0.0562,
|
| 1674 |
+
"step": 27100
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 8.52,
|
| 1678 |
+
"learning_rate": 7.78382053025153e-06,
|
| 1679 |
+
"loss": 0.0573,
|
| 1680 |
+
"step": 27200
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 8.55,
|
| 1684 |
+
"learning_rate": 7.762576478585996e-06,
|
| 1685 |
+
"loss": 0.0572,
|
| 1686 |
+
"step": 27300
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 8.59,
|
| 1690 |
+
"learning_rate": 7.741332426920463e-06,
|
| 1691 |
+
"loss": 0.0583,
|
| 1692 |
+
"step": 27400
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 8.62,
|
| 1696 |
+
"learning_rate": 7.72008837525493e-06,
|
| 1697 |
+
"loss": 0.0571,
|
| 1698 |
+
"step": 27500
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 8.65,
|
| 1702 |
+
"learning_rate": 7.698844323589395e-06,
|
| 1703 |
+
"loss": 0.0565,
|
| 1704 |
+
"step": 27600
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 8.68,
|
| 1708 |
+
"learning_rate": 7.677600271923862e-06,
|
| 1709 |
+
"loss": 0.0571,
|
| 1710 |
+
"step": 27700
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 8.71,
|
| 1714 |
+
"learning_rate": 7.656356220258327e-06,
|
| 1715 |
+
"loss": 0.0561,
|
| 1716 |
+
"step": 27800
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 8.74,
|
| 1720 |
+
"learning_rate": 7.635112168592794e-06,
|
| 1721 |
+
"loss": 0.0565,
|
| 1722 |
+
"step": 27900
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 8.77,
|
| 1726 |
+
"learning_rate": 7.613868116927261e-06,
|
| 1727 |
+
"loss": 0.0572,
|
| 1728 |
+
"step": 28000
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 8.8,
|
| 1732 |
+
"learning_rate": 7.5926240652617276e-06,
|
| 1733 |
+
"loss": 0.0568,
|
| 1734 |
+
"step": 28100
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 8.84,
|
| 1738 |
+
"learning_rate": 7.571380013596194e-06,
|
| 1739 |
+
"loss": 0.0568,
|
| 1740 |
+
"step": 28200
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 8.87,
|
| 1744 |
+
"learning_rate": 7.55013596193066e-06,
|
| 1745 |
+
"loss": 0.0569,
|
| 1746 |
+
"step": 28300
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 8.9,
|
| 1750 |
+
"learning_rate": 7.528891910265126e-06,
|
| 1751 |
+
"loss": 0.0564,
|
| 1752 |
+
"step": 28400
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 8.93,
|
| 1756 |
+
"learning_rate": 7.507647858599592e-06,
|
| 1757 |
+
"loss": 0.058,
|
| 1758 |
+
"step": 28500
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 8.96,
|
| 1762 |
+
"learning_rate": 7.4864038069340584e-06,
|
| 1763 |
+
"loss": 0.0568,
|
| 1764 |
+
"step": 28600
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 8.99,
|
| 1768 |
+
"learning_rate": 7.465159755268525e-06,
|
| 1769 |
+
"loss": 0.0577,
|
| 1770 |
+
"step": 28700
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 9.02,
|
| 1774 |
+
"learning_rate": 7.443915703602991e-06,
|
| 1775 |
+
"loss": 0.0543,
|
| 1776 |
+
"step": 28800
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 9.06,
|
| 1780 |
+
"learning_rate": 7.422671651937457e-06,
|
| 1781 |
+
"loss": 0.0535,
|
| 1782 |
+
"step": 28900
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 9.09,
|
| 1786 |
+
"learning_rate": 7.401427600271925e-06,
|
| 1787 |
+
"loss": 0.0532,
|
| 1788 |
+
"step": 29000
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 9.12,
|
| 1792 |
+
"learning_rate": 7.380183548606391e-06,
|
| 1793 |
+
"loss": 0.0527,
|
| 1794 |
+
"step": 29100
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 9.15,
|
| 1798 |
+
"learning_rate": 7.358939496940857e-06,
|
| 1799 |
+
"loss": 0.0539,
|
| 1800 |
+
"step": 29200
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 9.18,
|
| 1804 |
+
"learning_rate": 7.337695445275323e-06,
|
| 1805 |
+
"loss": 0.0519,
|
| 1806 |
+
"step": 29300
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 9.21,
|
| 1810 |
+
"learning_rate": 7.3164513936097895e-06,
|
| 1811 |
+
"loss": 0.0513,
|
| 1812 |
+
"step": 29400
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 9.24,
|
| 1816 |
+
"learning_rate": 7.295207341944256e-06,
|
| 1817 |
+
"loss": 0.0544,
|
| 1818 |
+
"step": 29500
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 9.27,
|
| 1822 |
+
"learning_rate": 7.273963290278722e-06,
|
| 1823 |
+
"loss": 0.0541,
|
| 1824 |
+
"step": 29600
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 9.31,
|
| 1828 |
+
"learning_rate": 7.252719238613188e-06,
|
| 1829 |
+
"loss": 0.0529,
|
| 1830 |
+
"step": 29700
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 9.34,
|
| 1834 |
+
"learning_rate": 7.231475186947654e-06,
|
| 1835 |
+
"loss": 0.0536,
|
| 1836 |
+
"step": 29800
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 9.37,
|
| 1840 |
+
"learning_rate": 7.210231135282121e-06,
|
| 1841 |
+
"loss": 0.0543,
|
| 1842 |
+
"step": 29900
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 9.4,
|
| 1846 |
+
"learning_rate": 7.1889870836165874e-06,
|
| 1847 |
+
"loss": 0.0523,
|
| 1848 |
+
"step": 30000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 9.4,
|
| 1852 |
+
"eval_loss": 0.16908282041549683,
|
| 1853 |
+
"eval_runtime": 112.1691,
|
| 1854 |
+
"eval_samples_per_second": 79.71,
|
| 1855 |
+
"eval_steps_per_second": 1.248,
|
| 1856 |
+
"step": 30000
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 9.43,
|
| 1860 |
+
"learning_rate": 7.1677430319510544e-06,
|
| 1861 |
+
"loss": 0.052,
|
| 1862 |
+
"step": 30100
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 9.46,
|
| 1866 |
+
"learning_rate": 7.146498980285521e-06,
|
| 1867 |
+
"loss": 0.0524,
|
| 1868 |
+
"step": 30200
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 9.49,
|
| 1872 |
+
"learning_rate": 7.125254928619987e-06,
|
| 1873 |
+
"loss": 0.0531,
|
| 1874 |
+
"step": 30300
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 9.53,
|
| 1878 |
+
"learning_rate": 7.104010876954453e-06,
|
| 1879 |
+
"loss": 0.0521,
|
| 1880 |
+
"step": 30400
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 9.56,
|
| 1884 |
+
"learning_rate": 7.082766825288919e-06,
|
| 1885 |
+
"loss": 0.0527,
|
| 1886 |
+
"step": 30500
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 9.59,
|
| 1890 |
+
"learning_rate": 7.061522773623386e-06,
|
| 1891 |
+
"loss": 0.0519,
|
| 1892 |
+
"step": 30600
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 9.62,
|
| 1896 |
+
"learning_rate": 7.040278721957852e-06,
|
| 1897 |
+
"loss": 0.0513,
|
| 1898 |
+
"step": 30700
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 9.65,
|
| 1902 |
+
"learning_rate": 7.0190346702923185e-06,
|
| 1903 |
+
"loss": 0.0524,
|
| 1904 |
+
"step": 30800
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 9.68,
|
| 1908 |
+
"learning_rate": 6.997790618626785e-06,
|
| 1909 |
+
"loss": 0.0543,
|
| 1910 |
+
"step": 30900
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 9.71,
|
| 1914 |
+
"learning_rate": 6.976546566961251e-06,
|
| 1915 |
+
"loss": 0.0539,
|
| 1916 |
+
"step": 31000
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 9.74,
|
| 1920 |
+
"learning_rate": 6.955302515295717e-06,
|
| 1921 |
+
"loss": 0.0512,
|
| 1922 |
+
"step": 31100
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 9.78,
|
| 1926 |
+
"learning_rate": 6.934058463630184e-06,
|
| 1927 |
+
"loss": 0.0514,
|
| 1928 |
+
"step": 31200
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 9.81,
|
| 1932 |
+
"learning_rate": 6.912814411964651e-06,
|
| 1933 |
+
"loss": 0.052,
|
| 1934 |
+
"step": 31300
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 9.84,
|
| 1938 |
+
"learning_rate": 6.891570360299117e-06,
|
| 1939 |
+
"loss": 0.0524,
|
| 1940 |
+
"step": 31400
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 9.87,
|
| 1944 |
+
"learning_rate": 6.8703263086335834e-06,
|
| 1945 |
+
"loss": 0.0525,
|
| 1946 |
+
"step": 31500
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 9.9,
|
| 1950 |
+
"learning_rate": 6.84908225696805e-06,
|
| 1951 |
+
"loss": 0.0526,
|
| 1952 |
+
"step": 31600
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 9.93,
|
| 1956 |
+
"learning_rate": 6.827838205302516e-06,
|
| 1957 |
+
"loss": 0.0523,
|
| 1958 |
+
"step": 31700
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 9.96,
|
| 1962 |
+
"learning_rate": 6.806594153636982e-06,
|
| 1963 |
+
"loss": 0.0527,
|
| 1964 |
+
"step": 31800
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 10.0,
|
| 1968 |
+
"learning_rate": 6.785350101971448e-06,
|
| 1969 |
+
"loss": 0.0525,
|
| 1970 |
+
"step": 31900
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 10.03,
|
| 1974 |
+
"learning_rate": 6.764106050305914e-06,
|
| 1975 |
+
"loss": 0.0482,
|
| 1976 |
+
"step": 32000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 10.06,
|
| 1980 |
+
"learning_rate": 6.7428619986403805e-06,
|
| 1981 |
+
"loss": 0.0487,
|
| 1982 |
+
"step": 32100
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 10.09,
|
| 1986 |
+
"learning_rate": 6.721617946974848e-06,
|
| 1987 |
+
"loss": 0.0497,
|
| 1988 |
+
"step": 32200
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 10.12,
|
| 1992 |
+
"learning_rate": 6.7003738953093145e-06,
|
| 1993 |
+
"loss": 0.0474,
|
| 1994 |
+
"step": 32300
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 10.15,
|
| 1998 |
+
"learning_rate": 6.679129843643781e-06,
|
| 1999 |
+
"loss": 0.0484,
|
| 2000 |
+
"step": 32400
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 10.18,
|
| 2004 |
+
"learning_rate": 6.657885791978247e-06,
|
| 2005 |
+
"loss": 0.0491,
|
| 2006 |
+
"step": 32500
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 10.21,
|
| 2010 |
+
"learning_rate": 6.636641740312713e-06,
|
| 2011 |
+
"loss": 0.0475,
|
| 2012 |
+
"step": 32600
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 10.25,
|
| 2016 |
+
"learning_rate": 6.615397688647179e-06,
|
| 2017 |
+
"loss": 0.049,
|
| 2018 |
+
"step": 32700
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 10.28,
|
| 2022 |
+
"learning_rate": 6.594153636981645e-06,
|
| 2023 |
+
"loss": 0.0477,
|
| 2024 |
+
"step": 32800
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 10.31,
|
| 2028 |
+
"learning_rate": 6.5729095853161116e-06,
|
| 2029 |
+
"loss": 0.0474,
|
| 2030 |
+
"step": 32900
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 10.34,
|
| 2034 |
+
"learning_rate": 6.551665533650578e-06,
|
| 2035 |
+
"loss": 0.0506,
|
| 2036 |
+
"step": 33000
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 10.37,
|
| 2040 |
+
"learning_rate": 6.530421481985044e-06,
|
| 2041 |
+
"loss": 0.0479,
|
| 2042 |
+
"step": 33100
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 10.4,
|
| 2046 |
+
"learning_rate": 6.50917743031951e-06,
|
| 2047 |
+
"loss": 0.0477,
|
| 2048 |
+
"step": 33200
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 10.43,
|
| 2052 |
+
"learning_rate": 6.487933378653978e-06,
|
| 2053 |
+
"loss": 0.0489,
|
| 2054 |
+
"step": 33300
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 10.47,
|
| 2058 |
+
"learning_rate": 6.466689326988444e-06,
|
| 2059 |
+
"loss": 0.0484,
|
| 2060 |
+
"step": 33400
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 10.5,
|
| 2064 |
+
"learning_rate": 6.44544527532291e-06,
|
| 2065 |
+
"loss": 0.0498,
|
| 2066 |
+
"step": 33500
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 10.53,
|
| 2070 |
+
"learning_rate": 6.4242012236573765e-06,
|
| 2071 |
+
"loss": 0.0485,
|
| 2072 |
+
"step": 33600
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 10.56,
|
| 2076 |
+
"learning_rate": 6.402957171991843e-06,
|
| 2077 |
+
"loss": 0.0484,
|
| 2078 |
+
"step": 33700
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 10.59,
|
| 2082 |
+
"learning_rate": 6.381713120326309e-06,
|
| 2083 |
+
"loss": 0.0468,
|
| 2084 |
+
"step": 33800
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 10.62,
|
| 2088 |
+
"learning_rate": 6.360469068660775e-06,
|
| 2089 |
+
"loss": 0.0483,
|
| 2090 |
+
"step": 33900
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 10.65,
|
| 2094 |
+
"learning_rate": 6.339225016995241e-06,
|
| 2095 |
+
"loss": 0.0493,
|
| 2096 |
+
"step": 34000
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 10.68,
|
| 2100 |
+
"learning_rate": 6.317980965329707e-06,
|
| 2101 |
+
"loss": 0.048,
|
| 2102 |
+
"step": 34100
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 10.72,
|
| 2106 |
+
"learning_rate": 6.2967369136641735e-06,
|
| 2107 |
+
"loss": 0.0505,
|
| 2108 |
+
"step": 34200
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 10.75,
|
| 2112 |
+
"learning_rate": 6.27549286199864e-06,
|
| 2113 |
+
"loss": 0.0487,
|
| 2114 |
+
"step": 34300
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 10.78,
|
| 2118 |
+
"learning_rate": 6.2542488103331076e-06,
|
| 2119 |
+
"loss": 0.0482,
|
| 2120 |
+
"step": 34400
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 10.81,
|
| 2124 |
+
"learning_rate": 6.233004758667573e-06,
|
| 2125 |
+
"loss": 0.0469,
|
| 2126 |
+
"step": 34500
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 10.84,
|
| 2130 |
+
"learning_rate": 6.21176070700204e-06,
|
| 2131 |
+
"loss": 0.0493,
|
| 2132 |
+
"step": 34600
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 10.87,
|
| 2136 |
+
"learning_rate": 6.190516655336506e-06,
|
| 2137 |
+
"loss": 0.0488,
|
| 2138 |
+
"step": 34700
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 10.9,
|
| 2142 |
+
"learning_rate": 6.169272603670972e-06,
|
| 2143 |
+
"loss": 0.0484,
|
| 2144 |
+
"step": 34800
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 10.94,
|
| 2148 |
+
"learning_rate": 6.1480285520054384e-06,
|
| 2149 |
+
"loss": 0.0478,
|
| 2150 |
+
"step": 34900
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 10.97,
|
| 2154 |
+
"learning_rate": 6.126784500339905e-06,
|
| 2155 |
+
"loss": 0.0497,
|
| 2156 |
+
"step": 35000
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 10.97,
|
| 2160 |
+
"eval_loss": 0.16288596391677856,
|
| 2161 |
+
"eval_runtime": 95.7573,
|
| 2162 |
+
"eval_samples_per_second": 93.372,
|
| 2163 |
+
"eval_steps_per_second": 1.462,
|
| 2164 |
+
"step": 35000
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 11.0,
|
| 2168 |
+
"learning_rate": 6.105540448674371e-06,
|
| 2169 |
+
"loss": 0.0481,
|
| 2170 |
+
"step": 35100
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 11.03,
|
| 2174 |
+
"learning_rate": 6.084296397008838e-06,
|
| 2175 |
+
"loss": 0.0456,
|
| 2176 |
+
"step": 35200
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 11.06,
|
| 2180 |
+
"learning_rate": 6.063052345343304e-06,
|
| 2181 |
+
"loss": 0.044,
|
| 2182 |
+
"step": 35300
|
| 2183 |
+
},
|
| 2184 |
+
{
|
| 2185 |
+
"epoch": 11.09,
|
| 2186 |
+
"learning_rate": 6.04180829367777e-06,
|
| 2187 |
+
"loss": 0.0443,
|
| 2188 |
+
"step": 35400
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 11.12,
|
| 2192 |
+
"learning_rate": 6.020564242012236e-06,
|
| 2193 |
+
"loss": 0.045,
|
| 2194 |
+
"step": 35500
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 11.15,
|
| 2198 |
+
"learning_rate": 5.9993201903467025e-06,
|
| 2199 |
+
"loss": 0.0435,
|
| 2200 |
+
"step": 35600
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 11.19,
|
| 2204 |
+
"learning_rate": 5.9780761386811695e-06,
|
| 2205 |
+
"loss": 0.0448,
|
| 2206 |
+
"step": 35700
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 11.22,
|
| 2210 |
+
"learning_rate": 5.956832087015636e-06,
|
| 2211 |
+
"loss": 0.0443,
|
| 2212 |
+
"step": 35800
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 11.25,
|
| 2216 |
+
"learning_rate": 5.935588035350102e-06,
|
| 2217 |
+
"loss": 0.0447,
|
| 2218 |
+
"step": 35900
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 11.28,
|
| 2222 |
+
"learning_rate": 5.914343983684569e-06,
|
| 2223 |
+
"loss": 0.0443,
|
| 2224 |
+
"step": 36000
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 11.31,
|
| 2228 |
+
"learning_rate": 5.893099932019035e-06,
|
| 2229 |
+
"loss": 0.0454,
|
| 2230 |
+
"step": 36100
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"epoch": 11.34,
|
| 2234 |
+
"learning_rate": 5.871855880353501e-06,
|
| 2235 |
+
"loss": 0.0439,
|
| 2236 |
+
"step": 36200
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 11.37,
|
| 2240 |
+
"learning_rate": 5.8506118286879674e-06,
|
| 2241 |
+
"loss": 0.0444,
|
| 2242 |
+
"step": 36300
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 11.41,
|
| 2246 |
+
"learning_rate": 5.8293677770224345e-06,
|
| 2247 |
+
"loss": 0.0453,
|
| 2248 |
+
"step": 36400
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 11.44,
|
| 2252 |
+
"learning_rate": 5.808123725356901e-06,
|
| 2253 |
+
"loss": 0.044,
|
| 2254 |
+
"step": 36500
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"epoch": 11.47,
|
| 2258 |
+
"learning_rate": 5.786879673691367e-06,
|
| 2259 |
+
"loss": 0.0449,
|
| 2260 |
+
"step": 36600
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 11.5,
|
| 2264 |
+
"learning_rate": 5.765635622025833e-06,
|
| 2265 |
+
"loss": 0.045,
|
| 2266 |
+
"step": 36700
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 11.53,
|
| 2270 |
+
"learning_rate": 5.7443915703603e-06,
|
| 2271 |
+
"loss": 0.0446,
|
| 2272 |
+
"step": 36800
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"epoch": 11.56,
|
| 2276 |
+
"learning_rate": 5.723147518694766e-06,
|
| 2277 |
+
"loss": 0.0453,
|
| 2278 |
+
"step": 36900
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 11.59,
|
| 2282 |
+
"learning_rate": 5.701903467029232e-06,
|
| 2283 |
+
"loss": 0.0439,
|
| 2284 |
+
"step": 37000
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 11.62,
|
| 2288 |
+
"learning_rate": 5.6806594153636985e-06,
|
| 2289 |
+
"loss": 0.0465,
|
| 2290 |
+
"step": 37100
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 11.66,
|
| 2294 |
+
"learning_rate": 5.659415363698165e-06,
|
| 2295 |
+
"loss": 0.0446,
|
| 2296 |
+
"step": 37200
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 11.69,
|
| 2300 |
+
"learning_rate": 5.638171312032632e-06,
|
| 2301 |
+
"loss": 0.0437,
|
| 2302 |
+
"step": 37300
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 11.72,
|
| 2306 |
+
"learning_rate": 5.616927260367098e-06,
|
| 2307 |
+
"loss": 0.0446,
|
| 2308 |
+
"step": 37400
|
| 2309 |
+
},
|
| 2310 |
+
{
|
| 2311 |
+
"epoch": 11.75,
|
| 2312 |
+
"learning_rate": 5.595683208701564e-06,
|
| 2313 |
+
"loss": 0.0442,
|
| 2314 |
+
"step": 37500
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 11.78,
|
| 2318 |
+
"learning_rate": 5.57443915703603e-06,
|
| 2319 |
+
"loss": 0.044,
|
| 2320 |
+
"step": 37600
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 11.81,
|
| 2324 |
+
"learning_rate": 5.553195105370496e-06,
|
| 2325 |
+
"loss": 0.0446,
|
| 2326 |
+
"step": 37700
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 11.84,
|
| 2330 |
+
"learning_rate": 5.5319510537049634e-06,
|
| 2331 |
+
"loss": 0.0443,
|
| 2332 |
+
"step": 37800
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 11.88,
|
| 2336 |
+
"learning_rate": 5.51070700203943e-06,
|
| 2337 |
+
"loss": 0.0447,
|
| 2338 |
+
"step": 37900
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 11.91,
|
| 2342 |
+
"learning_rate": 5.489462950373896e-06,
|
| 2343 |
+
"loss": 0.0446,
|
| 2344 |
+
"step": 38000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 11.94,
|
| 2348 |
+
"learning_rate": 5.468218898708362e-06,
|
| 2349 |
+
"loss": 0.0442,
|
| 2350 |
+
"step": 38100
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 11.97,
|
| 2354 |
+
"learning_rate": 5.446974847042828e-06,
|
| 2355 |
+
"loss": 0.0464,
|
| 2356 |
+
"step": 38200
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 12.0,
|
| 2360 |
+
"learning_rate": 5.425730795377294e-06,
|
| 2361 |
+
"loss": 0.0439,
|
| 2362 |
+
"step": 38300
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 12.03,
|
| 2366 |
+
"learning_rate": 5.404486743711761e-06,
|
| 2367 |
+
"loss": 0.0408,
|
| 2368 |
+
"step": 38400
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 12.06,
|
| 2372 |
+
"learning_rate": 5.3832426920462275e-06,
|
| 2373 |
+
"loss": 0.0407,
|
| 2374 |
+
"step": 38500
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 12.09,
|
| 2378 |
+
"learning_rate": 5.361998640380694e-06,
|
| 2379 |
+
"loss": 0.0411,
|
| 2380 |
+
"step": 38600
|
| 2381 |
+
},
|
| 2382 |
+
{
|
| 2383 |
+
"epoch": 12.13,
|
| 2384 |
+
"learning_rate": 5.34075458871516e-06,
|
| 2385 |
+
"loss": 0.0417,
|
| 2386 |
+
"step": 38700
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 12.16,
|
| 2390 |
+
"learning_rate": 5.319510537049626e-06,
|
| 2391 |
+
"loss": 0.042,
|
| 2392 |
+
"step": 38800
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 12.19,
|
| 2396 |
+
"learning_rate": 5.298266485384093e-06,
|
| 2397 |
+
"loss": 0.0405,
|
| 2398 |
+
"step": 38900
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 12.22,
|
| 2402 |
+
"learning_rate": 5.277022433718559e-06,
|
| 2403 |
+
"loss": 0.0414,
|
| 2404 |
+
"step": 39000
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 12.25,
|
| 2408 |
+
"learning_rate": 5.255778382053025e-06,
|
| 2409 |
+
"loss": 0.042,
|
| 2410 |
+
"step": 39100
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 12.28,
|
| 2414 |
+
"learning_rate": 5.2345343303874916e-06,
|
| 2415 |
+
"loss": 0.0417,
|
| 2416 |
+
"step": 39200
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 12.31,
|
| 2420 |
+
"learning_rate": 5.213290278721958e-06,
|
| 2421 |
+
"loss": 0.0426,
|
| 2422 |
+
"step": 39300
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 12.35,
|
| 2426 |
+
"learning_rate": 5.192046227056425e-06,
|
| 2427 |
+
"loss": 0.0403,
|
| 2428 |
+
"step": 39400
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"epoch": 12.38,
|
| 2432 |
+
"learning_rate": 5.170802175390891e-06,
|
| 2433 |
+
"loss": 0.0397,
|
| 2434 |
+
"step": 39500
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 12.41,
|
| 2438 |
+
"learning_rate": 5.149558123725357e-06,
|
| 2439 |
+
"loss": 0.0413,
|
| 2440 |
+
"step": 39600
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 12.44,
|
| 2444 |
+
"learning_rate": 5.128314072059823e-06,
|
| 2445 |
+
"loss": 0.0415,
|
| 2446 |
+
"step": 39700
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 12.47,
|
| 2450 |
+
"learning_rate": 5.1070700203942895e-06,
|
| 2451 |
+
"loss": 0.0403,
|
| 2452 |
+
"step": 39800
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 12.5,
|
| 2456 |
+
"learning_rate": 5.085825968728756e-06,
|
| 2457 |
+
"loss": 0.0403,
|
| 2458 |
+
"step": 39900
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 12.53,
|
| 2462 |
+
"learning_rate": 5.064581917063223e-06,
|
| 2463 |
+
"loss": 0.0407,
|
| 2464 |
+
"step": 40000
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"epoch": 12.53,
|
| 2468 |
+
"eval_loss": 0.16060428321361542,
|
| 2469 |
+
"eval_runtime": 83.676,
|
| 2470 |
+
"eval_samples_per_second": 106.853,
|
| 2471 |
+
"eval_steps_per_second": 1.673,
|
| 2472 |
+
"step": 40000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 12.56,
|
| 2476 |
+
"learning_rate": 5.043337865397689e-06,
|
| 2477 |
+
"loss": 0.0414,
|
| 2478 |
+
"step": 40100
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 12.6,
|
| 2482 |
+
"learning_rate": 5.022093813732155e-06,
|
| 2483 |
+
"loss": 0.0414,
|
| 2484 |
+
"step": 40200
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 12.63,
|
| 2488 |
+
"learning_rate": 5.000849762066621e-06,
|
| 2489 |
+
"loss": 0.0405,
|
| 2490 |
+
"step": 40300
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 12.66,
|
| 2494 |
+
"learning_rate": 4.979605710401087e-06,
|
| 2495 |
+
"loss": 0.0407,
|
| 2496 |
+
"step": 40400
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 12.69,
|
| 2500 |
+
"learning_rate": 4.958361658735554e-06,
|
| 2501 |
+
"loss": 0.0405,
|
| 2502 |
+
"step": 40500
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 12.72,
|
| 2506 |
+
"learning_rate": 4.9371176070700206e-06,
|
| 2507 |
+
"loss": 0.0402,
|
| 2508 |
+
"step": 40600
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 12.75,
|
| 2512 |
+
"learning_rate": 4.915873555404487e-06,
|
| 2513 |
+
"loss": 0.0416,
|
| 2514 |
+
"step": 40700
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 12.78,
|
| 2518 |
+
"learning_rate": 4.894629503738953e-06,
|
| 2519 |
+
"loss": 0.0413,
|
| 2520 |
+
"step": 40800
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"epoch": 12.82,
|
| 2524 |
+
"learning_rate": 4.873385452073419e-06,
|
| 2525 |
+
"loss": 0.0411,
|
| 2526 |
+
"step": 40900
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 12.85,
|
| 2530 |
+
"learning_rate": 4.852141400407886e-06,
|
| 2531 |
+
"loss": 0.0398,
|
| 2532 |
+
"step": 41000
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 12.88,
|
| 2536 |
+
"learning_rate": 4.830897348742352e-06,
|
| 2537 |
+
"loss": 0.0406,
|
| 2538 |
+
"step": 41100
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 12.91,
|
| 2542 |
+
"learning_rate": 4.8096532970768185e-06,
|
| 2543 |
+
"loss": 0.0409,
|
| 2544 |
+
"step": 41200
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 12.94,
|
| 2548 |
+
"learning_rate": 4.788409245411285e-06,
|
| 2549 |
+
"loss": 0.0419,
|
| 2550 |
+
"step": 41300
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 12.97,
|
| 2554 |
+
"learning_rate": 4.767165193745751e-06,
|
| 2555 |
+
"loss": 0.0402,
|
| 2556 |
+
"step": 41400
|
| 2557 |
+
},
|
| 2558 |
+
{
|
| 2559 |
+
"epoch": 13.0,
|
| 2560 |
+
"learning_rate": 4.745921142080218e-06,
|
| 2561 |
+
"loss": 0.0414,
|
| 2562 |
+
"step": 41500
|
| 2563 |
+
},
|
| 2564 |
+
{
|
| 2565 |
+
"epoch": 13.03,
|
| 2566 |
+
"learning_rate": 4.724677090414684e-06,
|
| 2567 |
+
"loss": 0.0384,
|
| 2568 |
+
"step": 41600
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 13.07,
|
| 2572 |
+
"learning_rate": 4.70343303874915e-06,
|
| 2573 |
+
"loss": 0.0383,
|
| 2574 |
+
"step": 41700
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 13.1,
|
| 2578 |
+
"learning_rate": 4.682188987083617e-06,
|
| 2579 |
+
"loss": 0.0375,
|
| 2580 |
+
"step": 41800
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 13.13,
|
| 2584 |
+
"learning_rate": 4.660944935418083e-06,
|
| 2585 |
+
"loss": 0.0378,
|
| 2586 |
+
"step": 41900
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 13.16,
|
| 2590 |
+
"learning_rate": 4.6397008837525495e-06,
|
| 2591 |
+
"loss": 0.0373,
|
| 2592 |
+
"step": 42000
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 13.19,
|
| 2596 |
+
"learning_rate": 4.618456832087016e-06,
|
| 2597 |
+
"loss": 0.0367,
|
| 2598 |
+
"step": 42100
|
| 2599 |
+
},
|
| 2600 |
+
{
|
| 2601 |
+
"epoch": 13.22,
|
| 2602 |
+
"learning_rate": 4.597212780421483e-06,
|
| 2603 |
+
"loss": 0.0393,
|
| 2604 |
+
"step": 42200
|
| 2605 |
+
},
|
| 2606 |
+
{
|
| 2607 |
+
"epoch": 13.25,
|
| 2608 |
+
"learning_rate": 4.575968728755949e-06,
|
| 2609 |
+
"loss": 0.0381,
|
| 2610 |
+
"step": 42300
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 13.29,
|
| 2614 |
+
"learning_rate": 4.554724677090415e-06,
|
| 2615 |
+
"loss": 0.0392,
|
| 2616 |
+
"step": 42400
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 13.32,
|
| 2620 |
+
"learning_rate": 4.533480625424881e-06,
|
| 2621 |
+
"loss": 0.0371,
|
| 2622 |
+
"step": 42500
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 13.35,
|
| 2626 |
+
"learning_rate": 4.512236573759348e-06,
|
| 2627 |
+
"loss": 0.0395,
|
| 2628 |
+
"step": 42600
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 13.38,
|
| 2632 |
+
"learning_rate": 4.4909925220938145e-06,
|
| 2633 |
+
"loss": 0.0383,
|
| 2634 |
+
"step": 42700
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 13.41,
|
| 2638 |
+
"learning_rate": 4.469748470428281e-06,
|
| 2639 |
+
"loss": 0.037,
|
| 2640 |
+
"step": 42800
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 13.44,
|
| 2644 |
+
"learning_rate": 4.448504418762747e-06,
|
| 2645 |
+
"loss": 0.0371,
|
| 2646 |
+
"step": 42900
|
| 2647 |
+
},
|
| 2648 |
+
{
|
| 2649 |
+
"epoch": 13.47,
|
| 2650 |
+
"learning_rate": 4.427260367097213e-06,
|
| 2651 |
+
"loss": 0.0369,
|
| 2652 |
+
"step": 43000
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 13.5,
|
| 2656 |
+
"learning_rate": 4.406016315431679e-06,
|
| 2657 |
+
"loss": 0.039,
|
| 2658 |
+
"step": 43100
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 13.54,
|
| 2662 |
+
"learning_rate": 4.384772263766146e-06,
|
| 2663 |
+
"loss": 0.0365,
|
| 2664 |
+
"step": 43200
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 13.57,
|
| 2668 |
+
"learning_rate": 4.363528212100612e-06,
|
| 2669 |
+
"loss": 0.0379,
|
| 2670 |
+
"step": 43300
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 13.6,
|
| 2674 |
+
"learning_rate": 4.3422841604350785e-06,
|
| 2675 |
+
"loss": 0.0387,
|
| 2676 |
+
"step": 43400
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 13.63,
|
| 2680 |
+
"learning_rate": 4.321040108769545e-06,
|
| 2681 |
+
"loss": 0.0383,
|
| 2682 |
+
"step": 43500
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 13.66,
|
| 2686 |
+
"learning_rate": 4.299796057104011e-06,
|
| 2687 |
+
"loss": 0.0375,
|
| 2688 |
+
"step": 43600
|
| 2689 |
+
},
|
| 2690 |
+
{
|
| 2691 |
+
"epoch": 13.69,
|
| 2692 |
+
"learning_rate": 4.278552005438478e-06,
|
| 2693 |
+
"loss": 0.0376,
|
| 2694 |
+
"step": 43700
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 13.72,
|
| 2698 |
+
"learning_rate": 4.257307953772944e-06,
|
| 2699 |
+
"loss": 0.0376,
|
| 2700 |
+
"step": 43800
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 13.76,
|
| 2704 |
+
"learning_rate": 4.23606390210741e-06,
|
| 2705 |
+
"loss": 0.0377,
|
| 2706 |
+
"step": 43900
|
| 2707 |
+
},
|
| 2708 |
+
{
|
| 2709 |
+
"epoch": 13.79,
|
| 2710 |
+
"learning_rate": 4.214819850441876e-06,
|
| 2711 |
+
"loss": 0.0379,
|
| 2712 |
+
"step": 44000
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 13.82,
|
| 2716 |
+
"learning_rate": 4.193575798776343e-06,
|
| 2717 |
+
"loss": 0.0377,
|
| 2718 |
+
"step": 44100
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 13.85,
|
| 2722 |
+
"learning_rate": 4.17233174711081e-06,
|
| 2723 |
+
"loss": 0.037,
|
| 2724 |
+
"step": 44200
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 13.88,
|
| 2728 |
+
"learning_rate": 4.151087695445276e-06,
|
| 2729 |
+
"loss": 0.0382,
|
| 2730 |
+
"step": 44300
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 13.91,
|
| 2734 |
+
"learning_rate": 4.129843643779742e-06,
|
| 2735 |
+
"loss": 0.0396,
|
| 2736 |
+
"step": 44400
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 13.94,
|
| 2740 |
+
"learning_rate": 4.108599592114208e-06,
|
| 2741 |
+
"loss": 0.0377,
|
| 2742 |
+
"step": 44500
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 13.97,
|
| 2746 |
+
"learning_rate": 4.087355540448674e-06,
|
| 2747 |
+
"loss": 0.0381,
|
| 2748 |
+
"step": 44600
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 14.01,
|
| 2752 |
+
"learning_rate": 4.0661114887831405e-06,
|
| 2753 |
+
"loss": 0.0375,
|
| 2754 |
+
"step": 44700
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 14.04,
|
| 2758 |
+
"learning_rate": 4.0448674371176075e-06,
|
| 2759 |
+
"loss": 0.0348,
|
| 2760 |
+
"step": 44800
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 14.07,
|
| 2764 |
+
"learning_rate": 4.023623385452074e-06,
|
| 2765 |
+
"loss": 0.0349,
|
| 2766 |
+
"step": 44900
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 14.1,
|
| 2770 |
+
"learning_rate": 4.00237933378654e-06,
|
| 2771 |
+
"loss": 0.0345,
|
| 2772 |
+
"step": 45000
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 14.1,
|
| 2776 |
+
"eval_loss": 0.15830282866954803,
|
| 2777 |
+
"eval_runtime": 116.727,
|
| 2778 |
+
"eval_samples_per_second": 76.598,
|
| 2779 |
+
"eval_steps_per_second": 1.199,
|
| 2780 |
+
"step": 45000
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 14.13,
|
| 2784 |
+
"learning_rate": 3.981135282121006e-06,
|
| 2785 |
+
"loss": 0.0347,
|
| 2786 |
+
"step": 45100
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 14.16,
|
| 2790 |
+
"learning_rate": 3.959891230455472e-06,
|
| 2791 |
+
"loss": 0.0354,
|
| 2792 |
+
"step": 45200
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 14.19,
|
| 2796 |
+
"learning_rate": 3.938647178789939e-06,
|
| 2797 |
+
"loss": 0.0349,
|
| 2798 |
+
"step": 45300
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 14.23,
|
| 2802 |
+
"learning_rate": 3.917403127124405e-06,
|
| 2803 |
+
"loss": 0.0346,
|
| 2804 |
+
"step": 45400
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 14.26,
|
| 2808 |
+
"learning_rate": 3.896159075458872e-06,
|
| 2809 |
+
"loss": 0.0352,
|
| 2810 |
+
"step": 45500
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 14.29,
|
| 2814 |
+
"learning_rate": 3.874915023793338e-06,
|
| 2815 |
+
"loss": 0.0349,
|
| 2816 |
+
"step": 45600
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 14.32,
|
| 2820 |
+
"learning_rate": 3.853670972127804e-06,
|
| 2821 |
+
"loss": 0.0358,
|
| 2822 |
+
"step": 45700
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 14.35,
|
| 2826 |
+
"learning_rate": 3.832426920462271e-06,
|
| 2827 |
+
"loss": 0.0351,
|
| 2828 |
+
"step": 45800
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 14.38,
|
| 2832 |
+
"learning_rate": 3.811182868796737e-06,
|
| 2833 |
+
"loss": 0.0344,
|
| 2834 |
+
"step": 45900
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 14.41,
|
| 2838 |
+
"learning_rate": 3.7899388171312033e-06,
|
| 2839 |
+
"loss": 0.0354,
|
| 2840 |
+
"step": 46000
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 14.44,
|
| 2844 |
+
"learning_rate": 3.76869476546567e-06,
|
| 2845 |
+
"loss": 0.0355,
|
| 2846 |
+
"step": 46100
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 14.48,
|
| 2850 |
+
"learning_rate": 3.747450713800136e-06,
|
| 2851 |
+
"loss": 0.0352,
|
| 2852 |
+
"step": 46200
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 14.51,
|
| 2856 |
+
"learning_rate": 3.7262066621346022e-06,
|
| 2857 |
+
"loss": 0.035,
|
| 2858 |
+
"step": 46300
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 14.54,
|
| 2862 |
+
"learning_rate": 3.7049626104690693e-06,
|
| 2863 |
+
"loss": 0.0349,
|
| 2864 |
+
"step": 46400
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 14.57,
|
| 2868 |
+
"learning_rate": 3.6837185588035354e-06,
|
| 2869 |
+
"loss": 0.0349,
|
| 2870 |
+
"step": 46500
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 14.6,
|
| 2874 |
+
"learning_rate": 3.6624745071380016e-06,
|
| 2875 |
+
"loss": 0.0333,
|
| 2876 |
+
"step": 46600
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 14.63,
|
| 2880 |
+
"learning_rate": 3.641230455472468e-06,
|
| 2881 |
+
"loss": 0.0331,
|
| 2882 |
+
"step": 46700
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 14.66,
|
| 2886 |
+
"learning_rate": 3.619986403806934e-06,
|
| 2887 |
+
"loss": 0.0351,
|
| 2888 |
+
"step": 46800
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 14.7,
|
| 2892 |
+
"learning_rate": 3.598742352141401e-06,
|
| 2893 |
+
"loss": 0.0335,
|
| 2894 |
+
"step": 46900
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 14.73,
|
| 2898 |
+
"learning_rate": 3.577498300475867e-06,
|
| 2899 |
+
"loss": 0.0349,
|
| 2900 |
+
"step": 47000
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 14.76,
|
| 2904 |
+
"learning_rate": 3.5562542488103333e-06,
|
| 2905 |
+
"loss": 0.0352,
|
| 2906 |
+
"step": 47100
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 14.79,
|
| 2910 |
+
"learning_rate": 3.5350101971447995e-06,
|
| 2911 |
+
"loss": 0.034,
|
| 2912 |
+
"step": 47200
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 14.82,
|
| 2916 |
+
"learning_rate": 3.5137661454792657e-06,
|
| 2917 |
+
"loss": 0.0348,
|
| 2918 |
+
"step": 47300
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 14.85,
|
| 2922 |
+
"learning_rate": 3.4925220938137327e-06,
|
| 2923 |
+
"loss": 0.0345,
|
| 2924 |
+
"step": 47400
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 14.88,
|
| 2928 |
+
"learning_rate": 3.471278042148199e-06,
|
| 2929 |
+
"loss": 0.0345,
|
| 2930 |
+
"step": 47500
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 14.91,
|
| 2934 |
+
"learning_rate": 3.450033990482665e-06,
|
| 2935 |
+
"loss": 0.0353,
|
| 2936 |
+
"step": 47600
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 14.95,
|
| 2940 |
+
"learning_rate": 3.4287899388171312e-06,
|
| 2941 |
+
"loss": 0.0336,
|
| 2942 |
+
"step": 47700
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 14.98,
|
| 2946 |
+
"learning_rate": 3.4075458871515974e-06,
|
| 2947 |
+
"loss": 0.035,
|
| 2948 |
+
"step": 47800
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 15.01,
|
| 2952 |
+
"learning_rate": 3.3863018354860636e-06,
|
| 2953 |
+
"loss": 0.0348,
|
| 2954 |
+
"step": 47900
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 15.04,
|
| 2958 |
+
"learning_rate": 3.3650577838205306e-06,
|
| 2959 |
+
"loss": 0.0323,
|
| 2960 |
+
"step": 48000
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 15.07,
|
| 2964 |
+
"learning_rate": 3.3438137321549968e-06,
|
| 2965 |
+
"loss": 0.0319,
|
| 2966 |
+
"step": 48100
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 15.1,
|
| 2970 |
+
"learning_rate": 3.322569680489463e-06,
|
| 2971 |
+
"loss": 0.0325,
|
| 2972 |
+
"step": 48200
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 15.13,
|
| 2976 |
+
"learning_rate": 3.301325628823929e-06,
|
| 2977 |
+
"loss": 0.0324,
|
| 2978 |
+
"step": 48300
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 15.17,
|
| 2982 |
+
"learning_rate": 3.2800815771583953e-06,
|
| 2983 |
+
"loss": 0.0337,
|
| 2984 |
+
"step": 48400
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 15.2,
|
| 2988 |
+
"learning_rate": 3.2588375254928623e-06,
|
| 2989 |
+
"loss": 0.0322,
|
| 2990 |
+
"step": 48500
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 15.23,
|
| 2994 |
+
"learning_rate": 3.2375934738273285e-06,
|
| 2995 |
+
"loss": 0.0321,
|
| 2996 |
+
"step": 48600
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 15.26,
|
| 3000 |
+
"learning_rate": 3.2163494221617947e-06,
|
| 3001 |
+
"loss": 0.0327,
|
| 3002 |
+
"step": 48700
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 15.29,
|
| 3006 |
+
"learning_rate": 3.1951053704962613e-06,
|
| 3007 |
+
"loss": 0.0322,
|
| 3008 |
+
"step": 48800
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 15.32,
|
| 3012 |
+
"learning_rate": 3.1738613188307274e-06,
|
| 3013 |
+
"loss": 0.0334,
|
| 3014 |
+
"step": 48900
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 15.35,
|
| 3018 |
+
"learning_rate": 3.152617267165194e-06,
|
| 3019 |
+
"loss": 0.0321,
|
| 3020 |
+
"step": 49000
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 15.38,
|
| 3024 |
+
"learning_rate": 3.1313732154996602e-06,
|
| 3025 |
+
"loss": 0.0311,
|
| 3026 |
+
"step": 49100
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 15.42,
|
| 3030 |
+
"learning_rate": 3.110129163834127e-06,
|
| 3031 |
+
"loss": 0.0322,
|
| 3032 |
+
"step": 49200
|
| 3033 |
+
},
|
| 3034 |
+
{
|
| 3035 |
+
"epoch": 15.45,
|
| 3036 |
+
"learning_rate": 3.088885112168593e-06,
|
| 3037 |
+
"loss": 0.0317,
|
| 3038 |
+
"step": 49300
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 15.48,
|
| 3042 |
+
"learning_rate": 3.0676410605030596e-06,
|
| 3043 |
+
"loss": 0.0325,
|
| 3044 |
+
"step": 49400
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 15.51,
|
| 3048 |
+
"learning_rate": 3.0463970088375258e-06,
|
| 3049 |
+
"loss": 0.0329,
|
| 3050 |
+
"step": 49500
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 15.54,
|
| 3054 |
+
"learning_rate": 3.025152957171992e-06,
|
| 3055 |
+
"loss": 0.032,
|
| 3056 |
+
"step": 49600
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 15.57,
|
| 3060 |
+
"learning_rate": 3.0039089055064585e-06,
|
| 3061 |
+
"loss": 0.031,
|
| 3062 |
+
"step": 49700
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 15.6,
|
| 3066 |
+
"learning_rate": 2.9826648538409247e-06,
|
| 3067 |
+
"loss": 0.0312,
|
| 3068 |
+
"step": 49800
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 15.64,
|
| 3072 |
+
"learning_rate": 2.961420802175391e-06,
|
| 3073 |
+
"loss": 0.0319,
|
| 3074 |
+
"step": 49900
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 15.67,
|
| 3078 |
+
"learning_rate": 2.9401767505098575e-06,
|
| 3079 |
+
"loss": 0.0307,
|
| 3080 |
+
"step": 50000
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 15.67,
|
| 3084 |
+
"eval_loss": 0.15686601400375366,
|
| 3085 |
+
"eval_runtime": 104.9673,
|
| 3086 |
+
"eval_samples_per_second": 85.179,
|
| 3087 |
+
"eval_steps_per_second": 1.334,
|
| 3088 |
+
"step": 50000
|
| 3089 |
+
},
|
| 3090 |
+
{
|
| 3091 |
+
"epoch": 15.7,
|
| 3092 |
+
"learning_rate": 2.9189326988443237e-06,
|
| 3093 |
+
"loss": 0.0317,
|
| 3094 |
+
"step": 50100
|
| 3095 |
+
},
|
| 3096 |
+
{
|
| 3097 |
+
"epoch": 15.73,
|
| 3098 |
+
"learning_rate": 2.8976886471787902e-06,
|
| 3099 |
+
"loss": 0.0312,
|
| 3100 |
+
"step": 50200
|
| 3101 |
+
},
|
| 3102 |
+
{
|
| 3103 |
+
"epoch": 15.76,
|
| 3104 |
+
"learning_rate": 2.8764445955132564e-06,
|
| 3105 |
+
"loss": 0.0305,
|
| 3106 |
+
"step": 50300
|
| 3107 |
+
},
|
| 3108 |
+
{
|
| 3109 |
+
"epoch": 15.79,
|
| 3110 |
+
"learning_rate": 2.8552005438477226e-06,
|
| 3111 |
+
"loss": 0.033,
|
| 3112 |
+
"step": 50400
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 15.82,
|
| 3116 |
+
"learning_rate": 2.833956492182189e-06,
|
| 3117 |
+
"loss": 0.0318,
|
| 3118 |
+
"step": 50500
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 15.85,
|
| 3122 |
+
"learning_rate": 2.8127124405166554e-06,
|
| 3123 |
+
"loss": 0.0312,
|
| 3124 |
+
"step": 50600
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 15.89,
|
| 3128 |
+
"learning_rate": 2.7914683888511215e-06,
|
| 3129 |
+
"loss": 0.0318,
|
| 3130 |
+
"step": 50700
|
| 3131 |
+
},
|
| 3132 |
+
{
|
| 3133 |
+
"epoch": 15.92,
|
| 3134 |
+
"learning_rate": 2.770224337185588e-06,
|
| 3135 |
+
"loss": 0.0314,
|
| 3136 |
+
"step": 50800
|
| 3137 |
+
},
|
| 3138 |
+
{
|
| 3139 |
+
"epoch": 15.95,
|
| 3140 |
+
"learning_rate": 2.7489802855200543e-06,
|
| 3141 |
+
"loss": 0.0328,
|
| 3142 |
+
"step": 50900
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 15.98,
|
| 3146 |
+
"learning_rate": 2.727736233854521e-06,
|
| 3147 |
+
"loss": 0.0325,
|
| 3148 |
+
"step": 51000
|
| 3149 |
+
},
|
| 3150 |
+
{
|
| 3151 |
+
"epoch": 16.01,
|
| 3152 |
+
"learning_rate": 2.706492182188987e-06,
|
| 3153 |
+
"loss": 0.0312,
|
| 3154 |
+
"step": 51100
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 16.04,
|
| 3158 |
+
"learning_rate": 2.6852481305234533e-06,
|
| 3159 |
+
"loss": 0.0311,
|
| 3160 |
+
"step": 51200
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 16.07,
|
| 3164 |
+
"learning_rate": 2.66400407885792e-06,
|
| 3165 |
+
"loss": 0.0298,
|
| 3166 |
+
"step": 51300
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 16.11,
|
| 3170 |
+
"learning_rate": 2.642760027192386e-06,
|
| 3171 |
+
"loss": 0.0297,
|
| 3172 |
+
"step": 51400
|
| 3173 |
+
},
|
| 3174 |
+
{
|
| 3175 |
+
"epoch": 16.14,
|
| 3176 |
+
"learning_rate": 2.6215159755268526e-06,
|
| 3177 |
+
"loss": 0.03,
|
| 3178 |
+
"step": 51500
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 16.17,
|
| 3182 |
+
"learning_rate": 2.600271923861319e-06,
|
| 3183 |
+
"loss": 0.0306,
|
| 3184 |
+
"step": 51600
|
| 3185 |
+
},
|
| 3186 |
+
{
|
| 3187 |
+
"epoch": 16.2,
|
| 3188 |
+
"learning_rate": 2.5790278721957854e-06,
|
| 3189 |
+
"loss": 0.0289,
|
| 3190 |
+
"step": 51700
|
| 3191 |
+
},
|
| 3192 |
+
{
|
| 3193 |
+
"epoch": 16.23,
|
| 3194 |
+
"learning_rate": 2.5577838205302516e-06,
|
| 3195 |
+
"loss": 0.0288,
|
| 3196 |
+
"step": 51800
|
| 3197 |
+
},
|
| 3198 |
+
{
|
| 3199 |
+
"epoch": 16.26,
|
| 3200 |
+
"learning_rate": 2.536539768864718e-06,
|
| 3201 |
+
"loss": 0.0303,
|
| 3202 |
+
"step": 51900
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 16.29,
|
| 3206 |
+
"learning_rate": 2.5152957171991844e-06,
|
| 3207 |
+
"loss": 0.0289,
|
| 3208 |
+
"step": 52000
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 16.32,
|
| 3212 |
+
"learning_rate": 2.494051665533651e-06,
|
| 3213 |
+
"loss": 0.0285,
|
| 3214 |
+
"step": 52100
|
| 3215 |
+
},
|
| 3216 |
+
{
|
| 3217 |
+
"epoch": 16.36,
|
| 3218 |
+
"learning_rate": 2.472807613868117e-06,
|
| 3219 |
+
"loss": 0.0303,
|
| 3220 |
+
"step": 52200
|
| 3221 |
+
},
|
| 3222 |
+
{
|
| 3223 |
+
"epoch": 16.39,
|
| 3224 |
+
"learning_rate": 2.4515635622025833e-06,
|
| 3225 |
+
"loss": 0.0296,
|
| 3226 |
+
"step": 52300
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 16.42,
|
| 3230 |
+
"learning_rate": 2.43031951053705e-06,
|
| 3231 |
+
"loss": 0.0301,
|
| 3232 |
+
"step": 52400
|
| 3233 |
+
},
|
| 3234 |
+
{
|
| 3235 |
+
"epoch": 16.45,
|
| 3236 |
+
"learning_rate": 2.409075458871516e-06,
|
| 3237 |
+
"loss": 0.0285,
|
| 3238 |
+
"step": 52500
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 16.48,
|
| 3242 |
+
"learning_rate": 2.3878314072059827e-06,
|
| 3243 |
+
"loss": 0.0289,
|
| 3244 |
+
"step": 52600
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 16.51,
|
| 3248 |
+
"learning_rate": 2.366587355540449e-06,
|
| 3249 |
+
"loss": 0.0313,
|
| 3250 |
+
"step": 52700
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 16.54,
|
| 3254 |
+
"learning_rate": 2.345343303874915e-06,
|
| 3255 |
+
"loss": 0.0307,
|
| 3256 |
+
"step": 52800
|
| 3257 |
+
},
|
| 3258 |
+
{
|
| 3259 |
+
"epoch": 16.58,
|
| 3260 |
+
"learning_rate": 2.3240992522093816e-06,
|
| 3261 |
+
"loss": 0.0293,
|
| 3262 |
+
"step": 52900
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 16.61,
|
| 3266 |
+
"learning_rate": 2.302855200543848e-06,
|
| 3267 |
+
"loss": 0.0307,
|
| 3268 |
+
"step": 53000
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 16.64,
|
| 3272 |
+
"learning_rate": 2.281611148878314e-06,
|
| 3273 |
+
"loss": 0.0296,
|
| 3274 |
+
"step": 53100
|
| 3275 |
+
},
|
| 3276 |
+
{
|
| 3277 |
+
"epoch": 16.67,
|
| 3278 |
+
"learning_rate": 2.2603670972127806e-06,
|
| 3279 |
+
"loss": 0.0295,
|
| 3280 |
+
"step": 53200
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 16.7,
|
| 3284 |
+
"learning_rate": 2.2391230455472467e-06,
|
| 3285 |
+
"loss": 0.0288,
|
| 3286 |
+
"step": 53300
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 16.73,
|
| 3290 |
+
"learning_rate": 2.2178789938817133e-06,
|
| 3291 |
+
"loss": 0.0299,
|
| 3292 |
+
"step": 53400
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 16.76,
|
| 3296 |
+
"learning_rate": 2.1966349422161795e-06,
|
| 3297 |
+
"loss": 0.029,
|
| 3298 |
+
"step": 53500
|
| 3299 |
+
},
|
| 3300 |
+
{
|
| 3301 |
+
"epoch": 16.79,
|
| 3302 |
+
"learning_rate": 2.1753908905506457e-06,
|
| 3303 |
+
"loss": 0.0291,
|
| 3304 |
+
"step": 53600
|
| 3305 |
+
},
|
| 3306 |
+
{
|
| 3307 |
+
"epoch": 16.83,
|
| 3308 |
+
"learning_rate": 2.1541468388851123e-06,
|
| 3309 |
+
"loss": 0.0286,
|
| 3310 |
+
"step": 53700
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 16.86,
|
| 3314 |
+
"learning_rate": 2.1329027872195785e-06,
|
| 3315 |
+
"loss": 0.0287,
|
| 3316 |
+
"step": 53800
|
| 3317 |
+
},
|
| 3318 |
+
{
|
| 3319 |
+
"epoch": 16.89,
|
| 3320 |
+
"learning_rate": 2.1116587355540446e-06,
|
| 3321 |
+
"loss": 0.0293,
|
| 3322 |
+
"step": 53900
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 16.92,
|
| 3326 |
+
"learning_rate": 2.0904146838885112e-06,
|
| 3327 |
+
"loss": 0.0292,
|
| 3328 |
+
"step": 54000
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 16.95,
|
| 3332 |
+
"learning_rate": 2.0691706322229774e-06,
|
| 3333 |
+
"loss": 0.0295,
|
| 3334 |
+
"step": 54100
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 16.98,
|
| 3338 |
+
"learning_rate": 2.047926580557444e-06,
|
| 3339 |
+
"loss": 0.0299,
|
| 3340 |
+
"step": 54200
|
| 3341 |
+
},
|
| 3342 |
+
{
|
| 3343 |
+
"epoch": 17.01,
|
| 3344 |
+
"learning_rate": 2.02668252889191e-06,
|
| 3345 |
+
"loss": 0.0287,
|
| 3346 |
+
"step": 54300
|
| 3347 |
+
},
|
| 3348 |
+
{
|
| 3349 |
+
"epoch": 17.05,
|
| 3350 |
+
"learning_rate": 2.0054384772263768e-06,
|
| 3351 |
+
"loss": 0.0292,
|
| 3352 |
+
"step": 54400
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 17.08,
|
| 3356 |
+
"learning_rate": 1.984194425560843e-06,
|
| 3357 |
+
"loss": 0.0272,
|
| 3358 |
+
"step": 54500
|
| 3359 |
+
},
|
| 3360 |
+
{
|
| 3361 |
+
"epoch": 17.11,
|
| 3362 |
+
"learning_rate": 1.9629503738953096e-06,
|
| 3363 |
+
"loss": 0.0271,
|
| 3364 |
+
"step": 54600
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 17.14,
|
| 3368 |
+
"learning_rate": 1.9417063222297757e-06,
|
| 3369 |
+
"loss": 0.0275,
|
| 3370 |
+
"step": 54700
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 17.17,
|
| 3374 |
+
"learning_rate": 1.9204622705642423e-06,
|
| 3375 |
+
"loss": 0.027,
|
| 3376 |
+
"step": 54800
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 17.2,
|
| 3380 |
+
"learning_rate": 1.8992182188987083e-06,
|
| 3381 |
+
"loss": 0.0282,
|
| 3382 |
+
"step": 54900
|
| 3383 |
+
},
|
| 3384 |
+
{
|
| 3385 |
+
"epoch": 17.23,
|
| 3386 |
+
"learning_rate": 1.8779741672331749e-06,
|
| 3387 |
+
"loss": 0.0266,
|
| 3388 |
+
"step": 55000
|
| 3389 |
+
},
|
| 3390 |
+
{
|
| 3391 |
+
"epoch": 17.23,
|
| 3392 |
+
"eval_loss": 0.15546093881130219,
|
| 3393 |
+
"eval_runtime": 80.0972,
|
| 3394 |
+
"eval_samples_per_second": 111.627,
|
| 3395 |
+
"eval_steps_per_second": 1.748,
|
| 3396 |
+
"step": 55000
|
| 3397 |
+
},
|
| 3398 |
+
{
|
| 3399 |
+
"epoch": 17.26,
|
| 3400 |
+
"learning_rate": 1.856730115567641e-06,
|
| 3401 |
+
"loss": 0.028,
|
| 3402 |
+
"step": 55100
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 17.3,
|
| 3406 |
+
"learning_rate": 1.8354860639021074e-06,
|
| 3407 |
+
"loss": 0.0274,
|
| 3408 |
+
"step": 55200
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 17.33,
|
| 3412 |
+
"learning_rate": 1.8142420122365738e-06,
|
| 3413 |
+
"loss": 0.0279,
|
| 3414 |
+
"step": 55300
|
| 3415 |
+
},
|
| 3416 |
+
{
|
| 3417 |
+
"epoch": 17.36,
|
| 3418 |
+
"learning_rate": 1.7929979605710402e-06,
|
| 3419 |
+
"loss": 0.0288,
|
| 3420 |
+
"step": 55400
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 17.39,
|
| 3424 |
+
"learning_rate": 1.7717539089055068e-06,
|
| 3425 |
+
"loss": 0.027,
|
| 3426 |
+
"step": 55500
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 17.42,
|
| 3430 |
+
"learning_rate": 1.750509857239973e-06,
|
| 3431 |
+
"loss": 0.0267,
|
| 3432 |
+
"step": 55600
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 17.45,
|
| 3436 |
+
"learning_rate": 1.7292658055744392e-06,
|
| 3437 |
+
"loss": 0.0288,
|
| 3438 |
+
"step": 55700
|
| 3439 |
+
},
|
| 3440 |
+
{
|
| 3441 |
+
"epoch": 17.48,
|
| 3442 |
+
"learning_rate": 1.7080217539089058e-06,
|
| 3443 |
+
"loss": 0.027,
|
| 3444 |
+
"step": 55800
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 17.52,
|
| 3448 |
+
"learning_rate": 1.686777702243372e-06,
|
| 3449 |
+
"loss": 0.027,
|
| 3450 |
+
"step": 55900
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 17.55,
|
| 3454 |
+
"learning_rate": 1.6655336505778381e-06,
|
| 3455 |
+
"loss": 0.0272,
|
| 3456 |
+
"step": 56000
|
| 3457 |
+
},
|
| 3458 |
+
{
|
| 3459 |
+
"epoch": 17.58,
|
| 3460 |
+
"learning_rate": 1.6442895989123047e-06,
|
| 3461 |
+
"loss": 0.0274,
|
| 3462 |
+
"step": 56100
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 17.61,
|
| 3466 |
+
"learning_rate": 1.6230455472467709e-06,
|
| 3467 |
+
"loss": 0.0272,
|
| 3468 |
+
"step": 56200
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 17.64,
|
| 3472 |
+
"learning_rate": 1.6018014955812375e-06,
|
| 3473 |
+
"loss": 0.0282,
|
| 3474 |
+
"step": 56300
|
| 3475 |
+
},
|
| 3476 |
+
{
|
| 3477 |
+
"epoch": 17.67,
|
| 3478 |
+
"learning_rate": 1.5805574439157037e-06,
|
| 3479 |
+
"loss": 0.0259,
|
| 3480 |
+
"step": 56400
|
| 3481 |
+
},
|
| 3482 |
+
{
|
| 3483 |
+
"epoch": 17.7,
|
| 3484 |
+
"learning_rate": 1.55931339225017e-06,
|
| 3485 |
+
"loss": 0.028,
|
| 3486 |
+
"step": 56500
|
| 3487 |
+
},
|
| 3488 |
+
{
|
| 3489 |
+
"epoch": 17.73,
|
| 3490 |
+
"learning_rate": 1.5380693405846364e-06,
|
| 3491 |
+
"loss": 0.0276,
|
| 3492 |
+
"step": 56600
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 17.77,
|
| 3496 |
+
"learning_rate": 1.5168252889191028e-06,
|
| 3497 |
+
"loss": 0.0277,
|
| 3498 |
+
"step": 56700
|
| 3499 |
+
},
|
| 3500 |
+
{
|
| 3501 |
+
"epoch": 17.8,
|
| 3502 |
+
"learning_rate": 1.4955812372535692e-06,
|
| 3503 |
+
"loss": 0.0263,
|
| 3504 |
+
"step": 56800
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 17.83,
|
| 3508 |
+
"learning_rate": 1.4743371855880356e-06,
|
| 3509 |
+
"loss": 0.0267,
|
| 3510 |
+
"step": 56900
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 17.86,
|
| 3514 |
+
"learning_rate": 1.4530931339225018e-06,
|
| 3515 |
+
"loss": 0.0275,
|
| 3516 |
+
"step": 57000
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 17.89,
|
| 3520 |
+
"learning_rate": 1.4318490822569681e-06,
|
| 3521 |
+
"loss": 0.027,
|
| 3522 |
+
"step": 57100
|
| 3523 |
+
},
|
| 3524 |
+
{
|
| 3525 |
+
"epoch": 17.92,
|
| 3526 |
+
"learning_rate": 1.4106050305914345e-06,
|
| 3527 |
+
"loss": 0.0268,
|
| 3528 |
+
"step": 57200
|
| 3529 |
+
},
|
| 3530 |
+
{
|
| 3531 |
+
"epoch": 17.95,
|
| 3532 |
+
"learning_rate": 1.389360978925901e-06,
|
| 3533 |
+
"loss": 0.0277,
|
| 3534 |
+
"step": 57300
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 17.99,
|
| 3538 |
+
"learning_rate": 1.368116927260367e-06,
|
| 3539 |
+
"loss": 0.0268,
|
| 3540 |
+
"step": 57400
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 18.02,
|
| 3544 |
+
"learning_rate": 1.3468728755948335e-06,
|
| 3545 |
+
"loss": 0.0268,
|
| 3546 |
+
"step": 57500
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 18.05,
|
| 3550 |
+
"learning_rate": 1.3256288239292999e-06,
|
| 3551 |
+
"loss": 0.0253,
|
| 3552 |
+
"step": 57600
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 18.08,
|
| 3556 |
+
"learning_rate": 1.3043847722637663e-06,
|
| 3557 |
+
"loss": 0.0253,
|
| 3558 |
+
"step": 57700
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 18.11,
|
| 3562 |
+
"learning_rate": 1.2831407205982324e-06,
|
| 3563 |
+
"loss": 0.026,
|
| 3564 |
+
"step": 57800
|
| 3565 |
+
},
|
| 3566 |
+
{
|
| 3567 |
+
"epoch": 18.14,
|
| 3568 |
+
"learning_rate": 1.2618966689326988e-06,
|
| 3569 |
+
"loss": 0.0255,
|
| 3570 |
+
"step": 57900
|
| 3571 |
+
},
|
| 3572 |
+
{
|
| 3573 |
+
"epoch": 18.17,
|
| 3574 |
+
"learning_rate": 1.2406526172671652e-06,
|
| 3575 |
+
"loss": 0.0271,
|
| 3576 |
+
"step": 58000
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 18.2,
|
| 3580 |
+
"learning_rate": 1.2194085656016316e-06,
|
| 3581 |
+
"loss": 0.0264,
|
| 3582 |
+
"step": 58100
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 18.24,
|
| 3586 |
+
"learning_rate": 1.198164513936098e-06,
|
| 3587 |
+
"loss": 0.0266,
|
| 3588 |
+
"step": 58200
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 18.27,
|
| 3592 |
+
"learning_rate": 1.1769204622705644e-06,
|
| 3593 |
+
"loss": 0.0269,
|
| 3594 |
+
"step": 58300
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 18.3,
|
| 3598 |
+
"learning_rate": 1.1556764106050307e-06,
|
| 3599 |
+
"loss": 0.0251,
|
| 3600 |
+
"step": 58400
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 18.33,
|
| 3604 |
+
"learning_rate": 1.1344323589394971e-06,
|
| 3605 |
+
"loss": 0.026,
|
| 3606 |
+
"step": 58500
|
| 3607 |
+
},
|
| 3608 |
+
{
|
| 3609 |
+
"epoch": 18.36,
|
| 3610 |
+
"learning_rate": 1.1131883072739633e-06,
|
| 3611 |
+
"loss": 0.0249,
|
| 3612 |
+
"step": 58600
|
| 3613 |
+
},
|
| 3614 |
+
{
|
| 3615 |
+
"epoch": 18.39,
|
| 3616 |
+
"learning_rate": 1.0919442556084297e-06,
|
| 3617 |
+
"loss": 0.0264,
|
| 3618 |
+
"step": 58700
|
| 3619 |
+
},
|
| 3620 |
+
{
|
| 3621 |
+
"epoch": 18.42,
|
| 3622 |
+
"learning_rate": 1.070700203942896e-06,
|
| 3623 |
+
"loss": 0.025,
|
| 3624 |
+
"step": 58800
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 18.46,
|
| 3628 |
+
"learning_rate": 1.0494561522773625e-06,
|
| 3629 |
+
"loss": 0.0259,
|
| 3630 |
+
"step": 58900
|
| 3631 |
+
},
|
| 3632 |
+
{
|
| 3633 |
+
"epoch": 18.49,
|
| 3634 |
+
"learning_rate": 1.0282121006118286e-06,
|
| 3635 |
+
"loss": 0.0266,
|
| 3636 |
+
"step": 59000
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 18.52,
|
| 3640 |
+
"learning_rate": 1.006968048946295e-06,
|
| 3641 |
+
"loss": 0.0251,
|
| 3642 |
+
"step": 59100
|
| 3643 |
+
},
|
| 3644 |
+
{
|
| 3645 |
+
"epoch": 18.55,
|
| 3646 |
+
"learning_rate": 9.857239972807614e-07,
|
| 3647 |
+
"loss": 0.0265,
|
| 3648 |
+
"step": 59200
|
| 3649 |
+
},
|
| 3650 |
+
{
|
| 3651 |
+
"epoch": 18.58,
|
| 3652 |
+
"learning_rate": 9.644799456152278e-07,
|
| 3653 |
+
"loss": 0.026,
|
| 3654 |
+
"step": 59300
|
| 3655 |
+
},
|
| 3656 |
+
{
|
| 3657 |
+
"epoch": 18.61,
|
| 3658 |
+
"learning_rate": 9.432358939496941e-07,
|
| 3659 |
+
"loss": 0.0254,
|
| 3660 |
+
"step": 59400
|
| 3661 |
+
},
|
| 3662 |
+
{
|
| 3663 |
+
"epoch": 18.64,
|
| 3664 |
+
"learning_rate": 9.219918422841605e-07,
|
| 3665 |
+
"loss": 0.0251,
|
| 3666 |
+
"step": 59500
|
| 3667 |
+
},
|
| 3668 |
+
{
|
| 3669 |
+
"epoch": 18.67,
|
| 3670 |
+
"learning_rate": 9.007477906186269e-07,
|
| 3671 |
+
"loss": 0.025,
|
| 3672 |
+
"step": 59600
|
| 3673 |
+
},
|
| 3674 |
+
{
|
| 3675 |
+
"epoch": 18.71,
|
| 3676 |
+
"learning_rate": 8.795037389530932e-07,
|
| 3677 |
+
"loss": 0.0247,
|
| 3678 |
+
"step": 59700
|
| 3679 |
+
},
|
| 3680 |
+
{
|
| 3681 |
+
"epoch": 18.74,
|
| 3682 |
+
"learning_rate": 8.582596872875595e-07,
|
| 3683 |
+
"loss": 0.0265,
|
| 3684 |
+
"step": 59800
|
| 3685 |
+
},
|
| 3686 |
+
{
|
| 3687 |
+
"epoch": 18.77,
|
| 3688 |
+
"learning_rate": 8.370156356220259e-07,
|
| 3689 |
+
"loss": 0.0259,
|
| 3690 |
+
"step": 59900
|
| 3691 |
+
},
|
| 3692 |
+
{
|
| 3693 |
+
"epoch": 18.8,
|
| 3694 |
+
"learning_rate": 8.157715839564923e-07,
|
| 3695 |
+
"loss": 0.0244,
|
| 3696 |
+
"step": 60000
|
| 3697 |
+
},
|
| 3698 |
+
{
|
| 3699 |
+
"epoch": 18.8,
|
| 3700 |
+
"eval_loss": 0.15402325987815857,
|
| 3701 |
+
"eval_runtime": 105.5951,
|
| 3702 |
+
"eval_samples_per_second": 84.673,
|
| 3703 |
+
"eval_steps_per_second": 1.326,
|
| 3704 |
+
"step": 60000
|
| 3705 |
+
},
|
| 3706 |
+
{
|
| 3707 |
+
"epoch": 18.83,
|
| 3708 |
+
"learning_rate": 7.945275322909587e-07,
|
| 3709 |
+
"loss": 0.0252,
|
| 3710 |
+
"step": 60100
|
| 3711 |
+
},
|
| 3712 |
+
{
|
| 3713 |
+
"epoch": 18.86,
|
| 3714 |
+
"learning_rate": 7.73283480625425e-07,
|
| 3715 |
+
"loss": 0.0258,
|
| 3716 |
+
"step": 60200
|
| 3717 |
+
},
|
| 3718 |
+
{
|
| 3719 |
+
"epoch": 18.89,
|
| 3720 |
+
"learning_rate": 7.520394289598912e-07,
|
| 3721 |
+
"loss": 0.0256,
|
| 3722 |
+
"step": 60300
|
| 3723 |
+
},
|
| 3724 |
+
{
|
| 3725 |
+
"epoch": 18.93,
|
| 3726 |
+
"learning_rate": 7.307953772943576e-07,
|
| 3727 |
+
"loss": 0.0253,
|
| 3728 |
+
"step": 60400
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 18.96,
|
| 3732 |
+
"learning_rate": 7.095513256288239e-07,
|
| 3733 |
+
"loss": 0.0243,
|
| 3734 |
+
"step": 60500
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 18.99,
|
| 3738 |
+
"learning_rate": 6.883072739632904e-07,
|
| 3739 |
+
"loss": 0.0246,
|
| 3740 |
+
"step": 60600
|
| 3741 |
+
},
|
| 3742 |
+
{
|
| 3743 |
+
"epoch": 19.02,
|
| 3744 |
+
"learning_rate": 6.670632222977567e-07,
|
| 3745 |
+
"loss": 0.0254,
|
| 3746 |
+
"step": 60700
|
| 3747 |
+
},
|
| 3748 |
+
{
|
| 3749 |
+
"epoch": 19.05,
|
| 3750 |
+
"learning_rate": 6.458191706322231e-07,
|
| 3751 |
+
"loss": 0.0248,
|
| 3752 |
+
"step": 60800
|
| 3753 |
+
},
|
| 3754 |
+
{
|
| 3755 |
+
"epoch": 19.08,
|
| 3756 |
+
"learning_rate": 6.245751189666893e-07,
|
| 3757 |
+
"loss": 0.0237,
|
| 3758 |
+
"step": 60900
|
| 3759 |
+
},
|
| 3760 |
+
{
|
| 3761 |
+
"epoch": 19.11,
|
| 3762 |
+
"learning_rate": 6.033310673011557e-07,
|
| 3763 |
+
"loss": 0.0243,
|
| 3764 |
+
"step": 61000
|
| 3765 |
+
},
|
| 3766 |
+
{
|
| 3767 |
+
"epoch": 19.14,
|
| 3768 |
+
"learning_rate": 5.82087015635622e-07,
|
| 3769 |
+
"loss": 0.0248,
|
| 3770 |
+
"step": 61100
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 19.18,
|
| 3774 |
+
"learning_rate": 5.608429639700884e-07,
|
| 3775 |
+
"loss": 0.0251,
|
| 3776 |
+
"step": 61200
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 19.21,
|
| 3780 |
+
"learning_rate": 5.395989123045548e-07,
|
| 3781 |
+
"loss": 0.0239,
|
| 3782 |
+
"step": 61300
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 19.24,
|
| 3786 |
+
"learning_rate": 5.183548606390212e-07,
|
| 3787 |
+
"loss": 0.0234,
|
| 3788 |
+
"step": 61400
|
| 3789 |
+
},
|
| 3790 |
+
{
|
| 3791 |
+
"epoch": 19.27,
|
| 3792 |
+
"learning_rate": 4.971108089734875e-07,
|
| 3793 |
+
"loss": 0.0255,
|
| 3794 |
+
"step": 61500
|
| 3795 |
+
},
|
| 3796 |
+
{
|
| 3797 |
+
"epoch": 19.3,
|
| 3798 |
+
"learning_rate": 4.7586675730795384e-07,
|
| 3799 |
+
"loss": 0.024,
|
| 3800 |
+
"step": 61600
|
| 3801 |
+
},
|
| 3802 |
+
{
|
| 3803 |
+
"epoch": 19.33,
|
| 3804 |
+
"learning_rate": 4.546227056424201e-07,
|
| 3805 |
+
"loss": 0.0243,
|
| 3806 |
+
"step": 61700
|
| 3807 |
+
},
|
| 3808 |
+
{
|
| 3809 |
+
"epoch": 19.36,
|
| 3810 |
+
"learning_rate": 4.333786539768865e-07,
|
| 3811 |
+
"loss": 0.0238,
|
| 3812 |
+
"step": 61800
|
| 3813 |
+
},
|
| 3814 |
+
{
|
| 3815 |
+
"epoch": 19.4,
|
| 3816 |
+
"learning_rate": 4.1213460231135284e-07,
|
| 3817 |
+
"loss": 0.025,
|
| 3818 |
+
"step": 61900
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 19.43,
|
| 3822 |
+
"learning_rate": 3.908905506458192e-07,
|
| 3823 |
+
"loss": 0.0238,
|
| 3824 |
+
"step": 62000
|
| 3825 |
+
},
|
| 3826 |
+
{
|
| 3827 |
+
"epoch": 19.46,
|
| 3828 |
+
"learning_rate": 3.6964649898028556e-07,
|
| 3829 |
+
"loss": 0.0239,
|
| 3830 |
+
"step": 62100
|
| 3831 |
+
},
|
| 3832 |
+
{
|
| 3833 |
+
"epoch": 19.49,
|
| 3834 |
+
"learning_rate": 3.484024473147519e-07,
|
| 3835 |
+
"loss": 0.0241,
|
| 3836 |
+
"step": 62200
|
| 3837 |
+
},
|
| 3838 |
+
{
|
| 3839 |
+
"epoch": 19.52,
|
| 3840 |
+
"learning_rate": 3.271583956492182e-07,
|
| 3841 |
+
"loss": 0.0241,
|
| 3842 |
+
"step": 62300
|
| 3843 |
+
},
|
| 3844 |
+
{
|
| 3845 |
+
"epoch": 19.55,
|
| 3846 |
+
"learning_rate": 3.0591434398368456e-07,
|
| 3847 |
+
"loss": 0.0252,
|
| 3848 |
+
"step": 62400
|
| 3849 |
+
},
|
| 3850 |
+
{
|
| 3851 |
+
"epoch": 19.58,
|
| 3852 |
+
"learning_rate": 2.8467029231815094e-07,
|
| 3853 |
+
"loss": 0.0242,
|
| 3854 |
+
"step": 62500
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 19.61,
|
| 3858 |
+
"learning_rate": 2.634262406526173e-07,
|
| 3859 |
+
"loss": 0.0237,
|
| 3860 |
+
"step": 62600
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 19.65,
|
| 3864 |
+
"learning_rate": 2.421821889870836e-07,
|
| 3865 |
+
"loss": 0.0237,
|
| 3866 |
+
"step": 62700
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 19.68,
|
| 3870 |
+
"learning_rate": 2.2093813732154997e-07,
|
| 3871 |
+
"loss": 0.0242,
|
| 3872 |
+
"step": 62800
|
| 3873 |
+
},
|
| 3874 |
+
{
|
| 3875 |
+
"epoch": 19.71,
|
| 3876 |
+
"learning_rate": 1.9969408565601633e-07,
|
| 3877 |
+
"loss": 0.0246,
|
| 3878 |
+
"step": 62900
|
| 3879 |
+
},
|
| 3880 |
+
{
|
| 3881 |
+
"epoch": 19.74,
|
| 3882 |
+
"learning_rate": 1.7845003399048266e-07,
|
| 3883 |
+
"loss": 0.0244,
|
| 3884 |
+
"step": 63000
|
| 3885 |
+
},
|
| 3886 |
+
{
|
| 3887 |
+
"epoch": 19.77,
|
| 3888 |
+
"learning_rate": 1.5720598232494902e-07,
|
| 3889 |
+
"loss": 0.0242,
|
| 3890 |
+
"step": 63100
|
| 3891 |
+
},
|
| 3892 |
+
{
|
| 3893 |
+
"epoch": 19.8,
|
| 3894 |
+
"learning_rate": 1.3596193065941536e-07,
|
| 3895 |
+
"loss": 0.0243,
|
| 3896 |
+
"step": 63200
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 19.83,
|
| 3900 |
+
"learning_rate": 1.1471787899388172e-07,
|
| 3901 |
+
"loss": 0.024,
|
| 3902 |
+
"step": 63300
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 19.87,
|
| 3906 |
+
"learning_rate": 9.347382732834808e-08,
|
| 3907 |
+
"loss": 0.0241,
|
| 3908 |
+
"step": 63400
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 19.9,
|
| 3912 |
+
"learning_rate": 7.222977566281442e-08,
|
| 3913 |
+
"loss": 0.0234,
|
| 3914 |
+
"step": 63500
|
| 3915 |
+
},
|
| 3916 |
+
{
|
| 3917 |
+
"epoch": 19.93,
|
| 3918 |
+
"learning_rate": 5.098572399728077e-08,
|
| 3919 |
+
"loss": 0.0238,
|
| 3920 |
+
"step": 63600
|
| 3921 |
+
},
|
| 3922 |
+
{
|
| 3923 |
+
"epoch": 19.96,
|
| 3924 |
+
"learning_rate": 2.9741672331747115e-08,
|
| 3925 |
+
"loss": 0.0238,
|
| 3926 |
+
"step": 63700
|
| 3927 |
+
},
|
| 3928 |
+
{
|
| 3929 |
+
"epoch": 19.99,
|
| 3930 |
+
"learning_rate": 8.49762066621346e-09,
|
| 3931 |
+
"loss": 0.0246,
|
| 3932 |
+
"step": 63800
|
| 3933 |
+
},
|
| 3934 |
+
{
|
| 3935 |
+
"epoch": 20.0,
|
| 3936 |
+
"step": 63840,
|
| 3937 |
+
"total_flos": 5.895453426057216e+20,
|
| 3938 |
+
"train_loss": 0.07169675377377292,
|
| 3939 |
+
"train_runtime": 490800.9612,
|
| 3940 |
+
"train_samples_per_second": 33.299,
|
| 3941 |
+
"train_steps_per_second": 0.13
|
| 3942 |
+
}
|
| 3943 |
+
],
|
| 3944 |
+
"max_steps": 63840,
|
| 3945 |
+
"num_train_epochs": 21,
|
| 3946 |
+
"total_flos": 5.895453426057216e+20,
|
| 3947 |
+
"trial_name": null,
|
| 3948 |
+
"trial_params": null
|
| 3949 |
+
}
|