Commit
·
43ddc1a
1
Parent(s):
bd5c19c
End of training
Browse files- README.md +1 -1
- all_results.json +14 -0
- eval_results.json +9 -0
- train_results.json +8 -0
- trainer_state.json +3328 -0
README.md
CHANGED
|
@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
This model is a fine-tuned version of [cardiffnlp/twitter-roberta-base-2019-90m](https://huggingface.co/cardiffnlp/twitter-roberta-base-2019-90m) on an unknown dataset.
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
-
- Loss: 2.
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
|
|
|
| 15 |
|
| 16 |
This model is a fine-tuned version of [cardiffnlp/twitter-roberta-base-2019-90m](https://huggingface.co/cardiffnlp/twitter-roberta-base-2019-90m) on an unknown dataset.
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
+
- Loss: 2.5190
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.83,
|
| 3 |
+
"eval_loss": 2.5190134048461914,
|
| 4 |
+
"eval_runtime": 231.2769,
|
| 5 |
+
"eval_samples": 205510,
|
| 6 |
+
"eval_samples_per_second": 888.588,
|
| 7 |
+
"eval_steps_per_second": 55.539,
|
| 8 |
+
"perplexity": 12.416340717163417,
|
| 9 |
+
"train_loss": 2.773840421549479,
|
| 10 |
+
"train_runtime": 219744.0708,
|
| 11 |
+
"train_samples": 3904699,
|
| 12 |
+
"train_samples_per_second": 174.749,
|
| 13 |
+
"train_steps_per_second": 10.922
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.83,
|
| 3 |
+
"eval_loss": 2.5190134048461914,
|
| 4 |
+
"eval_runtime": 231.2769,
|
| 5 |
+
"eval_samples": 205510,
|
| 6 |
+
"eval_samples_per_second": 888.588,
|
| 7 |
+
"eval_steps_per_second": 55.539,
|
| 8 |
+
"perplexity": 12.416340717163417
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.83,
|
| 3 |
+
"train_loss": 2.773840421549479,
|
| 4 |
+
"train_runtime": 219744.0708,
|
| 5 |
+
"train_samples": 3904699,
|
| 6 |
+
"train_samples_per_second": 174.749,
|
| 7 |
+
"train_steps_per_second": 10.922
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3328 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 2.5151257514953613,
|
| 3 |
+
"best_model_checkpoint": "./model_tweets_2020_Q1_50/checkpoint-128000",
|
| 4 |
+
"epoch": 9.834292176820574,
|
| 5 |
+
"eval_steps": 8000,
|
| 6 |
+
"global_step": 2400000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.03,
|
| 13 |
+
"eval_loss": 2.6128780841827393,
|
| 14 |
+
"eval_runtime": 220.0915,
|
| 15 |
+
"eval_samples_per_second": 933.748,
|
| 16 |
+
"eval_steps_per_second": 58.362,
|
| 17 |
+
"step": 8000
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.07,
|
| 21 |
+
"learning_rate": 9.939131159843243e-06,
|
| 22 |
+
"loss": 2.7516,
|
| 23 |
+
"step": 16000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.07,
|
| 27 |
+
"eval_loss": 2.560152769088745,
|
| 28 |
+
"eval_runtime": 222.3718,
|
| 29 |
+
"eval_samples_per_second": 924.173,
|
| 30 |
+
"eval_steps_per_second": 57.764,
|
| 31 |
+
"step": 16000
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.1,
|
| 35 |
+
"eval_loss": 2.561044216156006,
|
| 36 |
+
"eval_runtime": 219.7796,
|
| 37 |
+
"eval_samples_per_second": 935.073,
|
| 38 |
+
"eval_steps_per_second": 58.445,
|
| 39 |
+
"step": 24000
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.13,
|
| 43 |
+
"learning_rate": 9.872425581589261e-06,
|
| 44 |
+
"loss": 2.6478,
|
| 45 |
+
"step": 32000
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.13,
|
| 49 |
+
"eval_loss": 2.558018684387207,
|
| 50 |
+
"eval_runtime": 219.7961,
|
| 51 |
+
"eval_samples_per_second": 935.003,
|
| 52 |
+
"eval_steps_per_second": 58.441,
|
| 53 |
+
"step": 32000
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.16,
|
| 57 |
+
"eval_loss": 2.5498745441436768,
|
| 58 |
+
"eval_runtime": 221.2066,
|
| 59 |
+
"eval_samples_per_second": 929.041,
|
| 60 |
+
"eval_steps_per_second": 58.068,
|
| 61 |
+
"step": 40000
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.2,
|
| 65 |
+
"learning_rate": 9.80572000333528e-06,
|
| 66 |
+
"loss": 2.6344,
|
| 67 |
+
"step": 48000
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.2,
|
| 71 |
+
"eval_loss": 2.542255163192749,
|
| 72 |
+
"eval_runtime": 219.3858,
|
| 73 |
+
"eval_samples_per_second": 936.752,
|
| 74 |
+
"eval_steps_per_second": 58.55,
|
| 75 |
+
"step": 48000
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.23,
|
| 79 |
+
"eval_loss": 2.5431501865386963,
|
| 80 |
+
"eval_runtime": 220.781,
|
| 81 |
+
"eval_samples_per_second": 930.832,
|
| 82 |
+
"eval_steps_per_second": 58.18,
|
| 83 |
+
"step": 56000
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.26,
|
| 87 |
+
"learning_rate": 9.739014425081299e-06,
|
| 88 |
+
"loss": 2.6174,
|
| 89 |
+
"step": 64000
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.26,
|
| 93 |
+
"eval_loss": 2.550448417663574,
|
| 94 |
+
"eval_runtime": 220.5,
|
| 95 |
+
"eval_samples_per_second": 932.018,
|
| 96 |
+
"eval_steps_per_second": 58.254,
|
| 97 |
+
"step": 64000
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.3,
|
| 101 |
+
"eval_loss": 2.5348613262176514,
|
| 102 |
+
"eval_runtime": 220.0149,
|
| 103 |
+
"eval_samples_per_second": 934.073,
|
| 104 |
+
"eval_steps_per_second": 58.382,
|
| 105 |
+
"step": 72000
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.33,
|
| 109 |
+
"learning_rate": 9.672308846827316e-06,
|
| 110 |
+
"loss": 2.5993,
|
| 111 |
+
"step": 80000
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.33,
|
| 115 |
+
"eval_loss": 2.533210039138794,
|
| 116 |
+
"eval_runtime": 224.2726,
|
| 117 |
+
"eval_samples_per_second": 916.34,
|
| 118 |
+
"eval_steps_per_second": 57.274,
|
| 119 |
+
"step": 80000
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.36,
|
| 123 |
+
"eval_loss": 2.5342018604278564,
|
| 124 |
+
"eval_runtime": 221.6101,
|
| 125 |
+
"eval_samples_per_second": 927.349,
|
| 126 |
+
"eval_steps_per_second": 57.962,
|
| 127 |
+
"step": 88000
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 0.39,
|
| 131 |
+
"learning_rate": 9.605603268573334e-06,
|
| 132 |
+
"loss": 2.5948,
|
| 133 |
+
"step": 96000
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 0.39,
|
| 137 |
+
"eval_loss": 2.5296876430511475,
|
| 138 |
+
"eval_runtime": 221.2575,
|
| 139 |
+
"eval_samples_per_second": 928.827,
|
| 140 |
+
"eval_steps_per_second": 58.055,
|
| 141 |
+
"step": 96000
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.43,
|
| 145 |
+
"eval_loss": 2.5301403999328613,
|
| 146 |
+
"eval_runtime": 221.9648,
|
| 147 |
+
"eval_samples_per_second": 925.868,
|
| 148 |
+
"eval_steps_per_second": 57.87,
|
| 149 |
+
"step": 104000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.46,
|
| 153 |
+
"learning_rate": 9.538897690319354e-06,
|
| 154 |
+
"loss": 2.5846,
|
| 155 |
+
"step": 112000
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.46,
|
| 159 |
+
"eval_loss": 2.5199761390686035,
|
| 160 |
+
"eval_runtime": 221.7756,
|
| 161 |
+
"eval_samples_per_second": 926.658,
|
| 162 |
+
"eval_steps_per_second": 57.919,
|
| 163 |
+
"step": 112000
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.49,
|
| 167 |
+
"eval_loss": 2.5174739360809326,
|
| 168 |
+
"eval_runtime": 222.132,
|
| 169 |
+
"eval_samples_per_second": 925.171,
|
| 170 |
+
"eval_steps_per_second": 57.826,
|
| 171 |
+
"step": 120000
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.52,
|
| 175 |
+
"learning_rate": 9.472192112065373e-06,
|
| 176 |
+
"loss": 2.5774,
|
| 177 |
+
"step": 128000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.52,
|
| 181 |
+
"eval_loss": 2.5151257514953613,
|
| 182 |
+
"eval_runtime": 222.5971,
|
| 183 |
+
"eval_samples_per_second": 923.238,
|
| 184 |
+
"eval_steps_per_second": 57.705,
|
| 185 |
+
"step": 128000
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.56,
|
| 189 |
+
"eval_loss": 2.512612819671631,
|
| 190 |
+
"eval_runtime": 225.9428,
|
| 191 |
+
"eval_samples_per_second": 909.567,
|
| 192 |
+
"eval_steps_per_second": 56.851,
|
| 193 |
+
"step": 136000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.59,
|
| 197 |
+
"learning_rate": 9.405486533811392e-06,
|
| 198 |
+
"loss": 2.5584,
|
| 199 |
+
"step": 144000
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.59,
|
| 203 |
+
"eval_loss": 2.5006816387176514,
|
| 204 |
+
"eval_runtime": 225.1567,
|
| 205 |
+
"eval_samples_per_second": 912.742,
|
| 206 |
+
"eval_steps_per_second": 57.049,
|
| 207 |
+
"step": 144000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.62,
|
| 211 |
+
"eval_loss": 2.5149295330047607,
|
| 212 |
+
"eval_runtime": 222.6305,
|
| 213 |
+
"eval_samples_per_second": 923.099,
|
| 214 |
+
"eval_steps_per_second": 57.697,
|
| 215 |
+
"step": 152000
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.66,
|
| 219 |
+
"learning_rate": 9.338780955557409e-06,
|
| 220 |
+
"loss": 2.5578,
|
| 221 |
+
"step": 160000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.66,
|
| 225 |
+
"eval_loss": 2.519784688949585,
|
| 226 |
+
"eval_runtime": 224.6765,
|
| 227 |
+
"eval_samples_per_second": 914.693,
|
| 228 |
+
"eval_steps_per_second": 57.171,
|
| 229 |
+
"step": 160000
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.69,
|
| 233 |
+
"eval_loss": 2.5136947631835938,
|
| 234 |
+
"eval_runtime": 223.084,
|
| 235 |
+
"eval_samples_per_second": 921.223,
|
| 236 |
+
"eval_steps_per_second": 57.579,
|
| 237 |
+
"step": 168000
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.72,
|
| 241 |
+
"learning_rate": 9.272075377303427e-06,
|
| 242 |
+
"loss": 2.5699,
|
| 243 |
+
"step": 176000
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.72,
|
| 247 |
+
"eval_loss": 2.5189852714538574,
|
| 248 |
+
"eval_runtime": 221.4167,
|
| 249 |
+
"eval_samples_per_second": 928.159,
|
| 250 |
+
"eval_steps_per_second": 58.013,
|
| 251 |
+
"step": 176000
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.75,
|
| 255 |
+
"eval_loss": 2.5280351638793945,
|
| 256 |
+
"eval_runtime": 220.8999,
|
| 257 |
+
"eval_samples_per_second": 930.331,
|
| 258 |
+
"eval_steps_per_second": 58.149,
|
| 259 |
+
"step": 184000
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.79,
|
| 263 |
+
"learning_rate": 9.205369799049446e-06,
|
| 264 |
+
"loss": 2.5648,
|
| 265 |
+
"step": 192000
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.79,
|
| 269 |
+
"eval_loss": 2.5205190181732178,
|
| 270 |
+
"eval_runtime": 220.8432,
|
| 271 |
+
"eval_samples_per_second": 930.57,
|
| 272 |
+
"eval_steps_per_second": 58.163,
|
| 273 |
+
"step": 192000
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.82,
|
| 277 |
+
"eval_loss": 2.5266730785369873,
|
| 278 |
+
"eval_runtime": 223.4826,
|
| 279 |
+
"eval_samples_per_second": 919.579,
|
| 280 |
+
"eval_steps_per_second": 57.477,
|
| 281 |
+
"step": 200000
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.85,
|
| 285 |
+
"learning_rate": 9.138664220795464e-06,
|
| 286 |
+
"loss": 2.5701,
|
| 287 |
+
"step": 208000
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.85,
|
| 291 |
+
"eval_loss": 2.533360719680786,
|
| 292 |
+
"eval_runtime": 222.9517,
|
| 293 |
+
"eval_samples_per_second": 921.769,
|
| 294 |
+
"eval_steps_per_second": 57.613,
|
| 295 |
+
"step": 208000
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.89,
|
| 299 |
+
"eval_loss": 2.5359556674957275,
|
| 300 |
+
"eval_runtime": 223.0368,
|
| 301 |
+
"eval_samples_per_second": 921.418,
|
| 302 |
+
"eval_steps_per_second": 57.591,
|
| 303 |
+
"step": 216000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.92,
|
| 307 |
+
"learning_rate": 9.071958642541483e-06,
|
| 308 |
+
"loss": 2.5688,
|
| 309 |
+
"step": 224000
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.92,
|
| 313 |
+
"eval_loss": 2.5403811931610107,
|
| 314 |
+
"eval_runtime": 219.8534,
|
| 315 |
+
"eval_samples_per_second": 934.759,
|
| 316 |
+
"eval_steps_per_second": 58.425,
|
| 317 |
+
"step": 224000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.95,
|
| 321 |
+
"eval_loss": 2.536496162414551,
|
| 322 |
+
"eval_runtime": 222.4622,
|
| 323 |
+
"eval_samples_per_second": 923.797,
|
| 324 |
+
"eval_steps_per_second": 57.74,
|
| 325 |
+
"step": 232000
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.98,
|
| 329 |
+
"learning_rate": 9.005253064287502e-06,
|
| 330 |
+
"loss": 2.5788,
|
| 331 |
+
"step": 240000
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.98,
|
| 335 |
+
"eval_loss": 2.5484516620635986,
|
| 336 |
+
"eval_runtime": 221.8944,
|
| 337 |
+
"eval_samples_per_second": 926.161,
|
| 338 |
+
"eval_steps_per_second": 57.888,
|
| 339 |
+
"step": 240000
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.02,
|
| 343 |
+
"eval_loss": 2.5406737327575684,
|
| 344 |
+
"eval_runtime": 223.364,
|
| 345 |
+
"eval_samples_per_second": 920.068,
|
| 346 |
+
"eval_steps_per_second": 57.507,
|
| 347 |
+
"step": 248000
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 1.05,
|
| 351 |
+
"learning_rate": 8.93854748603352e-06,
|
| 352 |
+
"loss": 2.5857,
|
| 353 |
+
"step": 256000
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.05,
|
| 357 |
+
"eval_loss": 2.5444178581237793,
|
| 358 |
+
"eval_runtime": 222.6684,
|
| 359 |
+
"eval_samples_per_second": 922.942,
|
| 360 |
+
"eval_steps_per_second": 57.687,
|
| 361 |
+
"step": 256000
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 1.08,
|
| 365 |
+
"eval_loss": 2.55073881149292,
|
| 366 |
+
"eval_runtime": 221.5347,
|
| 367 |
+
"eval_samples_per_second": 927.665,
|
| 368 |
+
"eval_steps_per_second": 57.982,
|
| 369 |
+
"step": 264000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.11,
|
| 373 |
+
"learning_rate": 8.871841907779539e-06,
|
| 374 |
+
"loss": 2.576,
|
| 375 |
+
"step": 272000
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 1.11,
|
| 379 |
+
"eval_loss": 2.5567193031311035,
|
| 380 |
+
"eval_runtime": 221.584,
|
| 381 |
+
"eval_samples_per_second": 927.459,
|
| 382 |
+
"eval_steps_per_second": 57.969,
|
| 383 |
+
"step": 272000
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.15,
|
| 387 |
+
"eval_loss": 2.5561089515686035,
|
| 388 |
+
"eval_runtime": 221.035,
|
| 389 |
+
"eval_samples_per_second": 929.762,
|
| 390 |
+
"eval_steps_per_second": 58.113,
|
| 391 |
+
"step": 280000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 1.18,
|
| 395 |
+
"learning_rate": 8.805136329525557e-06,
|
| 396 |
+
"loss": 2.5919,
|
| 397 |
+
"step": 288000
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 1.18,
|
| 401 |
+
"eval_loss": 2.564077138900757,
|
| 402 |
+
"eval_runtime": 221.6286,
|
| 403 |
+
"eval_samples_per_second": 927.272,
|
| 404 |
+
"eval_steps_per_second": 57.957,
|
| 405 |
+
"step": 288000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.21,
|
| 409 |
+
"eval_loss": 2.5574235916137695,
|
| 410 |
+
"eval_runtime": 221.5082,
|
| 411 |
+
"eval_samples_per_second": 927.776,
|
| 412 |
+
"eval_steps_per_second": 57.989,
|
| 413 |
+
"step": 296000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.25,
|
| 417 |
+
"learning_rate": 8.738430751271576e-06,
|
| 418 |
+
"loss": 2.5893,
|
| 419 |
+
"step": 304000
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.25,
|
| 423 |
+
"eval_loss": 2.5562856197357178,
|
| 424 |
+
"eval_runtime": 221.2963,
|
| 425 |
+
"eval_samples_per_second": 928.664,
|
| 426 |
+
"eval_steps_per_second": 58.044,
|
| 427 |
+
"step": 304000
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 1.28,
|
| 431 |
+
"eval_loss": 2.5707902908325195,
|
| 432 |
+
"eval_runtime": 220.7861,
|
| 433 |
+
"eval_samples_per_second": 930.811,
|
| 434 |
+
"eval_steps_per_second": 58.178,
|
| 435 |
+
"step": 312000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 1.31,
|
| 439 |
+
"learning_rate": 8.671725173017595e-06,
|
| 440 |
+
"loss": 2.5896,
|
| 441 |
+
"step": 320000
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.31,
|
| 445 |
+
"eval_loss": 2.571282386779785,
|
| 446 |
+
"eval_runtime": 222.5864,
|
| 447 |
+
"eval_samples_per_second": 923.282,
|
| 448 |
+
"eval_steps_per_second": 57.708,
|
| 449 |
+
"step": 320000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.34,
|
| 453 |
+
"eval_loss": 2.5756185054779053,
|
| 454 |
+
"eval_runtime": 220.8371,
|
| 455 |
+
"eval_samples_per_second": 930.596,
|
| 456 |
+
"eval_steps_per_second": 58.165,
|
| 457 |
+
"step": 328000
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 1.38,
|
| 461 |
+
"learning_rate": 8.605019594763613e-06,
|
| 462 |
+
"loss": 2.6066,
|
| 463 |
+
"step": 336000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.38,
|
| 467 |
+
"eval_loss": 2.5867831707000732,
|
| 468 |
+
"eval_runtime": 221.6633,
|
| 469 |
+
"eval_samples_per_second": 927.127,
|
| 470 |
+
"eval_steps_per_second": 57.948,
|
| 471 |
+
"step": 336000
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 1.41,
|
| 475 |
+
"eval_loss": 2.5921335220336914,
|
| 476 |
+
"eval_runtime": 220.8265,
|
| 477 |
+
"eval_samples_per_second": 930.64,
|
| 478 |
+
"eval_steps_per_second": 58.168,
|
| 479 |
+
"step": 344000
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.44,
|
| 483 |
+
"learning_rate": 8.538314016509632e-06,
|
| 484 |
+
"loss": 2.6121,
|
| 485 |
+
"step": 352000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.44,
|
| 489 |
+
"eval_loss": 2.605332136154175,
|
| 490 |
+
"eval_runtime": 224.0349,
|
| 491 |
+
"eval_samples_per_second": 917.312,
|
| 492 |
+
"eval_steps_per_second": 57.335,
|
| 493 |
+
"step": 352000
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.48,
|
| 497 |
+
"eval_loss": 2.60457706451416,
|
| 498 |
+
"eval_runtime": 220.8455,
|
| 499 |
+
"eval_samples_per_second": 930.56,
|
| 500 |
+
"eval_steps_per_second": 58.163,
|
| 501 |
+
"step": 360000
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.51,
|
| 505 |
+
"learning_rate": 8.471608438255649e-06,
|
| 506 |
+
"loss": 2.6161,
|
| 507 |
+
"step": 368000
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.51,
|
| 511 |
+
"eval_loss": 2.5993847846984863,
|
| 512 |
+
"eval_runtime": 222.5248,
|
| 513 |
+
"eval_samples_per_second": 923.537,
|
| 514 |
+
"eval_steps_per_second": 57.724,
|
| 515 |
+
"step": 368000
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.54,
|
| 519 |
+
"eval_loss": 2.6035287380218506,
|
| 520 |
+
"eval_runtime": 221.0365,
|
| 521 |
+
"eval_samples_per_second": 929.756,
|
| 522 |
+
"eval_steps_per_second": 58.113,
|
| 523 |
+
"step": 376000
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 1.57,
|
| 527 |
+
"learning_rate": 8.404902860001667e-06,
|
| 528 |
+
"loss": 2.6313,
|
| 529 |
+
"step": 384000
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.57,
|
| 533 |
+
"eval_loss": 2.6119377613067627,
|
| 534 |
+
"eval_runtime": 221.8478,
|
| 535 |
+
"eval_samples_per_second": 926.356,
|
| 536 |
+
"eval_steps_per_second": 57.9,
|
| 537 |
+
"step": 384000
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 1.61,
|
| 541 |
+
"eval_loss": 2.601569175720215,
|
| 542 |
+
"eval_runtime": 221.1561,
|
| 543 |
+
"eval_samples_per_second": 929.253,
|
| 544 |
+
"eval_steps_per_second": 58.081,
|
| 545 |
+
"step": 392000
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.64,
|
| 549 |
+
"learning_rate": 8.338197281747686e-06,
|
| 550 |
+
"loss": 2.6342,
|
| 551 |
+
"step": 400000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 1.64,
|
| 555 |
+
"eval_loss": 2.62074875831604,
|
| 556 |
+
"eval_runtime": 222.2544,
|
| 557 |
+
"eval_samples_per_second": 924.661,
|
| 558 |
+
"eval_steps_per_second": 57.794,
|
| 559 |
+
"step": 400000
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 1.67,
|
| 563 |
+
"eval_loss": 2.623107433319092,
|
| 564 |
+
"eval_runtime": 221.8176,
|
| 565 |
+
"eval_samples_per_second": 926.482,
|
| 566 |
+
"eval_steps_per_second": 57.908,
|
| 567 |
+
"step": 408000
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.7,
|
| 571 |
+
"learning_rate": 8.271491703493705e-06,
|
| 572 |
+
"loss": 2.6358,
|
| 573 |
+
"step": 416000
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 1.7,
|
| 577 |
+
"eval_loss": 2.6287713050842285,
|
| 578 |
+
"eval_runtime": 224.4588,
|
| 579 |
+
"eval_samples_per_second": 915.58,
|
| 580 |
+
"eval_steps_per_second": 57.227,
|
| 581 |
+
"step": 416000
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 1.74,
|
| 585 |
+
"eval_loss": 2.633507490158081,
|
| 586 |
+
"eval_runtime": 222.6705,
|
| 587 |
+
"eval_samples_per_second": 922.933,
|
| 588 |
+
"eval_steps_per_second": 57.686,
|
| 589 |
+
"step": 424000
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.77,
|
| 593 |
+
"learning_rate": 8.204786125239725e-06,
|
| 594 |
+
"loss": 2.6463,
|
| 595 |
+
"step": 432000
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 1.77,
|
| 599 |
+
"eval_loss": 2.63563871383667,
|
| 600 |
+
"eval_runtime": 221.9208,
|
| 601 |
+
"eval_samples_per_second": 926.051,
|
| 602 |
+
"eval_steps_per_second": 57.881,
|
| 603 |
+
"step": 432000
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.8,
|
| 607 |
+
"eval_loss": 2.6429896354675293,
|
| 608 |
+
"eval_runtime": 222.0647,
|
| 609 |
+
"eval_samples_per_second": 925.451,
|
| 610 |
+
"eval_steps_per_second": 57.844,
|
| 611 |
+
"step": 440000
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.84,
|
| 615 |
+
"learning_rate": 8.138080546985743e-06,
|
| 616 |
+
"loss": 2.6561,
|
| 617 |
+
"step": 448000
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 1.84,
|
| 621 |
+
"eval_loss": 2.643951416015625,
|
| 622 |
+
"eval_runtime": 222.3078,
|
| 623 |
+
"eval_samples_per_second": 924.439,
|
| 624 |
+
"eval_steps_per_second": 57.78,
|
| 625 |
+
"step": 448000
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 1.87,
|
| 629 |
+
"eval_loss": 2.6444947719573975,
|
| 630 |
+
"eval_runtime": 222.0032,
|
| 631 |
+
"eval_samples_per_second": 925.707,
|
| 632 |
+
"eval_steps_per_second": 57.86,
|
| 633 |
+
"step": 456000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.9,
|
| 637 |
+
"learning_rate": 8.07137496873176e-06,
|
| 638 |
+
"loss": 2.6748,
|
| 639 |
+
"step": 464000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.9,
|
| 643 |
+
"eval_loss": 2.6506636142730713,
|
| 644 |
+
"eval_runtime": 223.5942,
|
| 645 |
+
"eval_samples_per_second": 919.121,
|
| 646 |
+
"eval_steps_per_second": 57.448,
|
| 647 |
+
"step": 464000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.93,
|
| 651 |
+
"eval_loss": 2.6547434329986572,
|
| 652 |
+
"eval_runtime": 221.7433,
|
| 653 |
+
"eval_samples_per_second": 926.792,
|
| 654 |
+
"eval_steps_per_second": 57.927,
|
| 655 |
+
"step": 472000
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.97,
|
| 659 |
+
"learning_rate": 8.004669390477779e-06,
|
| 660 |
+
"loss": 2.6847,
|
| 661 |
+
"step": 480000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.97,
|
| 665 |
+
"eval_loss": 2.664285898208618,
|
| 666 |
+
"eval_runtime": 221.7152,
|
| 667 |
+
"eval_samples_per_second": 926.91,
|
| 668 |
+
"eval_steps_per_second": 57.935,
|
| 669 |
+
"step": 480000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.0,
|
| 673 |
+
"eval_loss": 2.670330762863159,
|
| 674 |
+
"eval_runtime": 222.5951,
|
| 675 |
+
"eval_samples_per_second": 923.246,
|
| 676 |
+
"eval_steps_per_second": 57.706,
|
| 677 |
+
"step": 488000
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 2.03,
|
| 681 |
+
"learning_rate": 7.937963812223798e-06,
|
| 682 |
+
"loss": 2.6772,
|
| 683 |
+
"step": 496000
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 2.03,
|
| 687 |
+
"eval_loss": 2.674689292907715,
|
| 688 |
+
"eval_runtime": 222.9188,
|
| 689 |
+
"eval_samples_per_second": 921.905,
|
| 690 |
+
"eval_steps_per_second": 57.622,
|
| 691 |
+
"step": 496000
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 2.07,
|
| 695 |
+
"eval_loss": 2.6655595302581787,
|
| 696 |
+
"eval_runtime": 221.9863,
|
| 697 |
+
"eval_samples_per_second": 925.778,
|
| 698 |
+
"eval_steps_per_second": 57.864,
|
| 699 |
+
"step": 504000
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 2.1,
|
| 703 |
+
"learning_rate": 7.871258233969816e-06,
|
| 704 |
+
"loss": 2.6668,
|
| 705 |
+
"step": 512000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 2.1,
|
| 709 |
+
"eval_loss": 2.676820993423462,
|
| 710 |
+
"eval_runtime": 222.0249,
|
| 711 |
+
"eval_samples_per_second": 925.617,
|
| 712 |
+
"eval_steps_per_second": 57.854,
|
| 713 |
+
"step": 512000
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 2.13,
|
| 717 |
+
"eval_loss": 2.669207811355591,
|
| 718 |
+
"eval_runtime": 221.7932,
|
| 719 |
+
"eval_samples_per_second": 926.584,
|
| 720 |
+
"eval_steps_per_second": 57.914,
|
| 721 |
+
"step": 520000
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 2.16,
|
| 725 |
+
"learning_rate": 7.804552655715835e-06,
|
| 726 |
+
"loss": 2.6802,
|
| 727 |
+
"step": 528000
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 2.16,
|
| 731 |
+
"eval_loss": 2.6729602813720703,
|
| 732 |
+
"eval_runtime": 221.933,
|
| 733 |
+
"eval_samples_per_second": 926.0,
|
| 734 |
+
"eval_steps_per_second": 57.878,
|
| 735 |
+
"step": 528000
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.2,
|
| 739 |
+
"eval_loss": 2.67464017868042,
|
| 740 |
+
"eval_runtime": 222.2198,
|
| 741 |
+
"eval_samples_per_second": 924.805,
|
| 742 |
+
"eval_steps_per_second": 57.803,
|
| 743 |
+
"step": 536000
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 2.23,
|
| 747 |
+
"learning_rate": 7.737847077461853e-06,
|
| 748 |
+
"loss": 2.6856,
|
| 749 |
+
"step": 544000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 2.23,
|
| 753 |
+
"eval_loss": 2.678727865219116,
|
| 754 |
+
"eval_runtime": 222.413,
|
| 755 |
+
"eval_samples_per_second": 924.002,
|
| 756 |
+
"eval_steps_per_second": 57.753,
|
| 757 |
+
"step": 544000
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 2.26,
|
| 761 |
+
"eval_loss": 2.6777920722961426,
|
| 762 |
+
"eval_runtime": 222.6488,
|
| 763 |
+
"eval_samples_per_second": 923.023,
|
| 764 |
+
"eval_steps_per_second": 57.692,
|
| 765 |
+
"step": 552000
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.29,
|
| 769 |
+
"learning_rate": 7.671141499207872e-06,
|
| 770 |
+
"loss": 2.6874,
|
| 771 |
+
"step": 560000
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 2.29,
|
| 775 |
+
"eval_loss": 2.6909406185150146,
|
| 776 |
+
"eval_runtime": 221.9152,
|
| 777 |
+
"eval_samples_per_second": 926.074,
|
| 778 |
+
"eval_steps_per_second": 57.882,
|
| 779 |
+
"step": 560000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 2.33,
|
| 783 |
+
"eval_loss": 2.691913604736328,
|
| 784 |
+
"eval_runtime": 222.3607,
|
| 785 |
+
"eval_samples_per_second": 924.219,
|
| 786 |
+
"eval_steps_per_second": 57.766,
|
| 787 |
+
"step": 568000
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 2.36,
|
| 791 |
+
"learning_rate": 7.604435920953891e-06,
|
| 792 |
+
"loss": 2.6956,
|
| 793 |
+
"step": 576000
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 2.36,
|
| 797 |
+
"eval_loss": 2.69474720954895,
|
| 798 |
+
"eval_runtime": 222.1948,
|
| 799 |
+
"eval_samples_per_second": 924.909,
|
| 800 |
+
"eval_steps_per_second": 57.81,
|
| 801 |
+
"step": 576000
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.39,
|
| 805 |
+
"eval_loss": 2.7031822204589844,
|
| 806 |
+
"eval_runtime": 223.2718,
|
| 807 |
+
"eval_samples_per_second": 920.447,
|
| 808 |
+
"eval_steps_per_second": 57.531,
|
| 809 |
+
"step": 584000
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 2.43,
|
| 813 |
+
"learning_rate": 7.537730342699909e-06,
|
| 814 |
+
"loss": 2.7081,
|
| 815 |
+
"step": 592000
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 2.43,
|
| 819 |
+
"eval_loss": 2.707876205444336,
|
| 820 |
+
"eval_runtime": 222.7056,
|
| 821 |
+
"eval_samples_per_second": 922.788,
|
| 822 |
+
"eval_steps_per_second": 57.677,
|
| 823 |
+
"step": 592000
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 2.46,
|
| 827 |
+
"eval_loss": 2.710313320159912,
|
| 828 |
+
"eval_runtime": 223.6841,
|
| 829 |
+
"eval_samples_per_second": 918.751,
|
| 830 |
+
"eval_steps_per_second": 57.425,
|
| 831 |
+
"step": 600000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 2.49,
|
| 835 |
+
"learning_rate": 7.471024764445928e-06,
|
| 836 |
+
"loss": 2.7124,
|
| 837 |
+
"step": 608000
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.49,
|
| 841 |
+
"eval_loss": 2.7138588428497314,
|
| 842 |
+
"eval_runtime": 222.1167,
|
| 843 |
+
"eval_samples_per_second": 925.234,
|
| 844 |
+
"eval_steps_per_second": 57.83,
|
| 845 |
+
"step": 608000
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 2.52,
|
| 849 |
+
"eval_loss": 2.710862636566162,
|
| 850 |
+
"eval_runtime": 222.0676,
|
| 851 |
+
"eval_samples_per_second": 925.439,
|
| 852 |
+
"eval_steps_per_second": 57.843,
|
| 853 |
+
"step": 616000
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 2.56,
|
| 857 |
+
"learning_rate": 7.4043191861919465e-06,
|
| 858 |
+
"loss": 2.7221,
|
| 859 |
+
"step": 624000
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 2.56,
|
| 863 |
+
"eval_loss": 2.7152557373046875,
|
| 864 |
+
"eval_runtime": 223.9434,
|
| 865 |
+
"eval_samples_per_second": 917.687,
|
| 866 |
+
"eval_steps_per_second": 57.358,
|
| 867 |
+
"step": 624000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.59,
|
| 871 |
+
"eval_loss": 2.735896348953247,
|
| 872 |
+
"eval_runtime": 222.2379,
|
| 873 |
+
"eval_samples_per_second": 924.73,
|
| 874 |
+
"eval_steps_per_second": 57.798,
|
| 875 |
+
"step": 632000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 2.62,
|
| 879 |
+
"learning_rate": 7.337613607937964e-06,
|
| 880 |
+
"loss": 2.7131,
|
| 881 |
+
"step": 640000
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.62,
|
| 885 |
+
"eval_loss": 2.7278735637664795,
|
| 886 |
+
"eval_runtime": 222.8278,
|
| 887 |
+
"eval_samples_per_second": 922.282,
|
| 888 |
+
"eval_steps_per_second": 57.645,
|
| 889 |
+
"step": 640000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 2.66,
|
| 893 |
+
"eval_loss": 2.737804651260376,
|
| 894 |
+
"eval_runtime": 223.5516,
|
| 895 |
+
"eval_samples_per_second": 919.296,
|
| 896 |
+
"eval_steps_per_second": 57.459,
|
| 897 |
+
"step": 648000
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 2.69,
|
| 901 |
+
"learning_rate": 7.270908029683983e-06,
|
| 902 |
+
"loss": 2.7268,
|
| 903 |
+
"step": 656000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 2.69,
|
| 907 |
+
"eval_loss": 2.73801851272583,
|
| 908 |
+
"eval_runtime": 222.415,
|
| 909 |
+
"eval_samples_per_second": 923.993,
|
| 910 |
+
"eval_steps_per_second": 57.752,
|
| 911 |
+
"step": 656000
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 2.72,
|
| 915 |
+
"eval_loss": 2.727534294128418,
|
| 916 |
+
"eval_runtime": 222.4068,
|
| 917 |
+
"eval_samples_per_second": 924.027,
|
| 918 |
+
"eval_steps_per_second": 57.755,
|
| 919 |
+
"step": 664000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 2.75,
|
| 923 |
+
"learning_rate": 7.2042024514300015e-06,
|
| 924 |
+
"loss": 2.7373,
|
| 925 |
+
"step": 672000
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 2.75,
|
| 929 |
+
"eval_loss": 2.7439777851104736,
|
| 930 |
+
"eval_runtime": 224.2928,
|
| 931 |
+
"eval_samples_per_second": 916.258,
|
| 932 |
+
"eval_steps_per_second": 57.269,
|
| 933 |
+
"step": 672000
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 2.79,
|
| 937 |
+
"eval_loss": 2.7381510734558105,
|
| 938 |
+
"eval_runtime": 223.3597,
|
| 939 |
+
"eval_samples_per_second": 920.086,
|
| 940 |
+
"eval_steps_per_second": 57.508,
|
| 941 |
+
"step": 680000
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.82,
|
| 945 |
+
"learning_rate": 7.13749687317602e-06,
|
| 946 |
+
"loss": 2.7576,
|
| 947 |
+
"step": 688000
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 2.82,
|
| 951 |
+
"eval_loss": 2.743011236190796,
|
| 952 |
+
"eval_runtime": 223.7189,
|
| 953 |
+
"eval_samples_per_second": 918.608,
|
| 954 |
+
"eval_steps_per_second": 57.416,
|
| 955 |
+
"step": 688000
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.85,
|
| 959 |
+
"eval_loss": 2.742142915725708,
|
| 960 |
+
"eval_runtime": 223.6634,
|
| 961 |
+
"eval_samples_per_second": 918.836,
|
| 962 |
+
"eval_steps_per_second": 57.43,
|
| 963 |
+
"step": 696000
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.88,
|
| 967 |
+
"learning_rate": 7.070791294922038e-06,
|
| 968 |
+
"loss": 2.7495,
|
| 969 |
+
"step": 704000
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 2.88,
|
| 973 |
+
"eval_loss": 2.751878261566162,
|
| 974 |
+
"eval_runtime": 223.6777,
|
| 975 |
+
"eval_samples_per_second": 918.777,
|
| 976 |
+
"eval_steps_per_second": 57.426,
|
| 977 |
+
"step": 704000
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 2.92,
|
| 981 |
+
"eval_loss": 2.749375104904175,
|
| 982 |
+
"eval_runtime": 223.9511,
|
| 983 |
+
"eval_samples_per_second": 917.656,
|
| 984 |
+
"eval_steps_per_second": 57.356,
|
| 985 |
+
"step": 712000
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 2.95,
|
| 989 |
+
"learning_rate": 7.0040857166680564e-06,
|
| 990 |
+
"loss": 2.7626,
|
| 991 |
+
"step": 720000
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 2.95,
|
| 995 |
+
"eval_loss": 2.754082202911377,
|
| 996 |
+
"eval_runtime": 222.8179,
|
| 997 |
+
"eval_samples_per_second": 922.323,
|
| 998 |
+
"eval_steps_per_second": 57.648,
|
| 999 |
+
"step": 720000
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 2.98,
|
| 1003 |
+
"eval_loss": 2.7496798038482666,
|
| 1004 |
+
"eval_runtime": 223.5475,
|
| 1005 |
+
"eval_samples_per_second": 919.312,
|
| 1006 |
+
"eval_steps_per_second": 57.46,
|
| 1007 |
+
"step": 728000
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 3.02,
|
| 1011 |
+
"learning_rate": 6.937380138414076e-06,
|
| 1012 |
+
"loss": 2.7551,
|
| 1013 |
+
"step": 736000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 3.02,
|
| 1017 |
+
"eval_loss": 2.762470245361328,
|
| 1018 |
+
"eval_runtime": 223.2771,
|
| 1019 |
+
"eval_samples_per_second": 920.426,
|
| 1020 |
+
"eval_steps_per_second": 57.529,
|
| 1021 |
+
"step": 736000
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 3.05,
|
| 1025 |
+
"eval_loss": 2.764946937561035,
|
| 1026 |
+
"eval_runtime": 222.8366,
|
| 1027 |
+
"eval_samples_per_second": 922.245,
|
| 1028 |
+
"eval_steps_per_second": 57.643,
|
| 1029 |
+
"step": 744000
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 3.08,
|
| 1033 |
+
"learning_rate": 6.8706745601600945e-06,
|
| 1034 |
+
"loss": 2.7606,
|
| 1035 |
+
"step": 752000
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 3.08,
|
| 1039 |
+
"eval_loss": 2.759011745452881,
|
| 1040 |
+
"eval_runtime": 223.0437,
|
| 1041 |
+
"eval_samples_per_second": 921.389,
|
| 1042 |
+
"eval_steps_per_second": 57.59,
|
| 1043 |
+
"step": 752000
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 3.11,
|
| 1047 |
+
"eval_loss": 2.7598087787628174,
|
| 1048 |
+
"eval_runtime": 223.0646,
|
| 1049 |
+
"eval_samples_per_second": 921.303,
|
| 1050 |
+
"eval_steps_per_second": 57.584,
|
| 1051 |
+
"step": 760000
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 3.15,
|
| 1055 |
+
"learning_rate": 6.803968981906113e-06,
|
| 1056 |
+
"loss": 2.7709,
|
| 1057 |
+
"step": 768000
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 3.15,
|
| 1061 |
+
"eval_loss": 2.7685601711273193,
|
| 1062 |
+
"eval_runtime": 223.0947,
|
| 1063 |
+
"eval_samples_per_second": 921.178,
|
| 1064 |
+
"eval_steps_per_second": 57.576,
|
| 1065 |
+
"step": 768000
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 3.18,
|
| 1069 |
+
"eval_loss": 2.769714117050171,
|
| 1070 |
+
"eval_runtime": 224.3593,
|
| 1071 |
+
"eval_samples_per_second": 915.986,
|
| 1072 |
+
"eval_steps_per_second": 57.252,
|
| 1073 |
+
"step": 776000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 3.21,
|
| 1077 |
+
"learning_rate": 6.737263403652131e-06,
|
| 1078 |
+
"loss": 2.7687,
|
| 1079 |
+
"step": 784000
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 3.21,
|
| 1083 |
+
"eval_loss": 2.7807157039642334,
|
| 1084 |
+
"eval_runtime": 224.1624,
|
| 1085 |
+
"eval_samples_per_second": 916.791,
|
| 1086 |
+
"eval_steps_per_second": 57.302,
|
| 1087 |
+
"step": 784000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 3.25,
|
| 1091 |
+
"eval_loss": 2.783041000366211,
|
| 1092 |
+
"eval_runtime": 224.6759,
|
| 1093 |
+
"eval_samples_per_second": 914.695,
|
| 1094 |
+
"eval_steps_per_second": 57.171,
|
| 1095 |
+
"step": 792000
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 3.28,
|
| 1099 |
+
"learning_rate": 6.6705578253981495e-06,
|
| 1100 |
+
"loss": 2.7745,
|
| 1101 |
+
"step": 800000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 3.28,
|
| 1105 |
+
"eval_loss": 2.7677245140075684,
|
| 1106 |
+
"eval_runtime": 223.5451,
|
| 1107 |
+
"eval_samples_per_second": 919.322,
|
| 1108 |
+
"eval_steps_per_second": 57.46,
|
| 1109 |
+
"step": 800000
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 3.31,
|
| 1113 |
+
"eval_loss": 2.789714813232422,
|
| 1114 |
+
"eval_runtime": 224.1358,
|
| 1115 |
+
"eval_samples_per_second": 916.9,
|
| 1116 |
+
"eval_steps_per_second": 57.309,
|
| 1117 |
+
"step": 808000
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 3.34,
|
| 1121 |
+
"learning_rate": 6.603852247144168e-06,
|
| 1122 |
+
"loss": 2.7596,
|
| 1123 |
+
"step": 816000
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 3.34,
|
| 1127 |
+
"eval_loss": 2.782496690750122,
|
| 1128 |
+
"eval_runtime": 224.2552,
|
| 1129 |
+
"eval_samples_per_second": 916.411,
|
| 1130 |
+
"eval_steps_per_second": 57.278,
|
| 1131 |
+
"step": 816000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 3.38,
|
| 1135 |
+
"eval_loss": 2.782914400100708,
|
| 1136 |
+
"eval_runtime": 225.4775,
|
| 1137 |
+
"eval_samples_per_second": 911.444,
|
| 1138 |
+
"eval_steps_per_second": 56.968,
|
| 1139 |
+
"step": 824000
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 3.41,
|
| 1143 |
+
"learning_rate": 6.537146668890187e-06,
|
| 1144 |
+
"loss": 2.7749,
|
| 1145 |
+
"step": 832000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 3.41,
|
| 1149 |
+
"eval_loss": 2.7868270874023438,
|
| 1150 |
+
"eval_runtime": 224.6633,
|
| 1151 |
+
"eval_samples_per_second": 914.747,
|
| 1152 |
+
"eval_steps_per_second": 57.174,
|
| 1153 |
+
"step": 832000
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 3.44,
|
| 1157 |
+
"eval_loss": 2.7875137329101562,
|
| 1158 |
+
"eval_runtime": 223.5713,
|
| 1159 |
+
"eval_samples_per_second": 919.214,
|
| 1160 |
+
"eval_steps_per_second": 57.454,
|
| 1161 |
+
"step": 840000
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 3.47,
|
| 1165 |
+
"learning_rate": 6.4704410906362044e-06,
|
| 1166 |
+
"loss": 2.7788,
|
| 1167 |
+
"step": 848000
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 3.47,
|
| 1171 |
+
"eval_loss": 2.7906622886657715,
|
| 1172 |
+
"eval_runtime": 223.506,
|
| 1173 |
+
"eval_samples_per_second": 919.483,
|
| 1174 |
+
"eval_steps_per_second": 57.471,
|
| 1175 |
+
"step": 848000
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 3.51,
|
| 1179 |
+
"eval_loss": 2.7915995121002197,
|
| 1180 |
+
"eval_runtime": 225.1506,
|
| 1181 |
+
"eval_samples_per_second": 912.767,
|
| 1182 |
+
"eval_steps_per_second": 57.051,
|
| 1183 |
+
"step": 856000
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 3.54,
|
| 1187 |
+
"learning_rate": 6.403735512382223e-06,
|
| 1188 |
+
"loss": 2.7792,
|
| 1189 |
+
"step": 864000
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 3.54,
|
| 1193 |
+
"eval_loss": 2.793468713760376,
|
| 1194 |
+
"eval_runtime": 223.5626,
|
| 1195 |
+
"eval_samples_per_second": 919.25,
|
| 1196 |
+
"eval_steps_per_second": 57.456,
|
| 1197 |
+
"step": 864000
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 3.57,
|
| 1201 |
+
"eval_loss": 2.789273977279663,
|
| 1202 |
+
"eval_runtime": 224.3088,
|
| 1203 |
+
"eval_samples_per_second": 916.192,
|
| 1204 |
+
"eval_steps_per_second": 57.265,
|
| 1205 |
+
"step": 872000
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 3.61,
|
| 1209 |
+
"learning_rate": 6.337029934128242e-06,
|
| 1210 |
+
"loss": 2.7871,
|
| 1211 |
+
"step": 880000
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 3.61,
|
| 1215 |
+
"eval_loss": 2.8002703189849854,
|
| 1216 |
+
"eval_runtime": 223.9319,
|
| 1217 |
+
"eval_samples_per_second": 917.734,
|
| 1218 |
+
"eval_steps_per_second": 57.361,
|
| 1219 |
+
"step": 880000
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 3.64,
|
| 1223 |
+
"eval_loss": 2.79728364944458,
|
| 1224 |
+
"eval_runtime": 224.811,
|
| 1225 |
+
"eval_samples_per_second": 914.145,
|
| 1226 |
+
"eval_steps_per_second": 57.137,
|
| 1227 |
+
"step": 888000
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 3.67,
|
| 1231 |
+
"learning_rate": 6.270324355874261e-06,
|
| 1232 |
+
"loss": 2.7967,
|
| 1233 |
+
"step": 896000
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 3.67,
|
| 1237 |
+
"eval_loss": 2.810227394104004,
|
| 1238 |
+
"eval_runtime": 223.8856,
|
| 1239 |
+
"eval_samples_per_second": 917.924,
|
| 1240 |
+
"eval_steps_per_second": 57.373,
|
| 1241 |
+
"step": 896000
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 3.7,
|
| 1245 |
+
"eval_loss": 2.811846971511841,
|
| 1246 |
+
"eval_runtime": 226.2552,
|
| 1247 |
+
"eval_samples_per_second": 908.31,
|
| 1248 |
+
"eval_steps_per_second": 56.772,
|
| 1249 |
+
"step": 904000
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 3.74,
|
| 1253 |
+
"learning_rate": 6.20361877762028e-06,
|
| 1254 |
+
"loss": 2.7896,
|
| 1255 |
+
"step": 912000
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 3.74,
|
| 1259 |
+
"eval_loss": 2.805860757827759,
|
| 1260 |
+
"eval_runtime": 224.7634,
|
| 1261 |
+
"eval_samples_per_second": 914.339,
|
| 1262 |
+
"eval_steps_per_second": 57.149,
|
| 1263 |
+
"step": 912000
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 3.77,
|
| 1267 |
+
"eval_loss": 2.8134682178497314,
|
| 1268 |
+
"eval_runtime": 223.9747,
|
| 1269 |
+
"eval_samples_per_second": 917.559,
|
| 1270 |
+
"eval_steps_per_second": 57.35,
|
| 1271 |
+
"step": 920000
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 3.8,
|
| 1275 |
+
"learning_rate": 6.1369131993662975e-06,
|
| 1276 |
+
"loss": 2.8021,
|
| 1277 |
+
"step": 928000
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 3.8,
|
| 1281 |
+
"eval_loss": 2.810849905014038,
|
| 1282 |
+
"eval_runtime": 226.3005,
|
| 1283 |
+
"eval_samples_per_second": 908.129,
|
| 1284 |
+
"eval_steps_per_second": 56.761,
|
| 1285 |
+
"step": 928000
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 3.84,
|
| 1289 |
+
"eval_loss": 2.8164422512054443,
|
| 1290 |
+
"eval_runtime": 227.0284,
|
| 1291 |
+
"eval_samples_per_second": 905.217,
|
| 1292 |
+
"eval_steps_per_second": 56.579,
|
| 1293 |
+
"step": 936000
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 3.87,
|
| 1297 |
+
"learning_rate": 6.070207621112316e-06,
|
| 1298 |
+
"loss": 2.7931,
|
| 1299 |
+
"step": 944000
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 3.87,
|
| 1303 |
+
"eval_loss": 2.8173489570617676,
|
| 1304 |
+
"eval_runtime": 225.5541,
|
| 1305 |
+
"eval_samples_per_second": 911.134,
|
| 1306 |
+
"eval_steps_per_second": 56.949,
|
| 1307 |
+
"step": 944000
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 3.9,
|
| 1311 |
+
"eval_loss": 2.829514741897583,
|
| 1312 |
+
"eval_runtime": 225.3778,
|
| 1313 |
+
"eval_samples_per_second": 911.847,
|
| 1314 |
+
"eval_steps_per_second": 56.993,
|
| 1315 |
+
"step": 952000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 3.93,
|
| 1319 |
+
"learning_rate": 6.003502042858335e-06,
|
| 1320 |
+
"loss": 2.8105,
|
| 1321 |
+
"step": 960000
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 3.93,
|
| 1325 |
+
"eval_loss": 2.8248379230499268,
|
| 1326 |
+
"eval_runtime": 224.9909,
|
| 1327 |
+
"eval_samples_per_second": 913.415,
|
| 1328 |
+
"eval_steps_per_second": 57.091,
|
| 1329 |
+
"step": 960000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 3.97,
|
| 1333 |
+
"eval_loss": 2.812260627746582,
|
| 1334 |
+
"eval_runtime": 225.6651,
|
| 1335 |
+
"eval_samples_per_second": 910.686,
|
| 1336 |
+
"eval_steps_per_second": 56.921,
|
| 1337 |
+
"step": 968000
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 4.0,
|
| 1341 |
+
"learning_rate": 5.936796464604353e-06,
|
| 1342 |
+
"loss": 2.805,
|
| 1343 |
+
"step": 976000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 4.0,
|
| 1347 |
+
"eval_loss": 2.829453468322754,
|
| 1348 |
+
"eval_runtime": 224.9048,
|
| 1349 |
+
"eval_samples_per_second": 913.764,
|
| 1350 |
+
"eval_steps_per_second": 57.113,
|
| 1351 |
+
"step": 976000
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 4.03,
|
| 1355 |
+
"eval_loss": 2.8286798000335693,
|
| 1356 |
+
"eval_runtime": 224.572,
|
| 1357 |
+
"eval_samples_per_second": 915.119,
|
| 1358 |
+
"eval_steps_per_second": 57.198,
|
| 1359 |
+
"step": 984000
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 4.06,
|
| 1363 |
+
"learning_rate": 5.870090886350371e-06,
|
| 1364 |
+
"loss": 2.7959,
|
| 1365 |
+
"step": 992000
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 4.06,
|
| 1369 |
+
"eval_loss": 2.826554298400879,
|
| 1370 |
+
"eval_runtime": 224.6548,
|
| 1371 |
+
"eval_samples_per_second": 914.781,
|
| 1372 |
+
"eval_steps_per_second": 57.177,
|
| 1373 |
+
"step": 992000
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 4.1,
|
| 1377 |
+
"eval_loss": 2.844482898712158,
|
| 1378 |
+
"eval_runtime": 225.9544,
|
| 1379 |
+
"eval_samples_per_second": 909.52,
|
| 1380 |
+
"eval_steps_per_second": 56.848,
|
| 1381 |
+
"step": 1000000
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 4.13,
|
| 1385 |
+
"learning_rate": 5.80338530809639e-06,
|
| 1386 |
+
"loss": 2.8126,
|
| 1387 |
+
"step": 1008000
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 4.13,
|
| 1391 |
+
"eval_loss": 2.8259875774383545,
|
| 1392 |
+
"eval_runtime": 226.7424,
|
| 1393 |
+
"eval_samples_per_second": 906.359,
|
| 1394 |
+
"eval_steps_per_second": 56.65,
|
| 1395 |
+
"step": 1008000
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 4.16,
|
| 1399 |
+
"eval_loss": 2.829375743865967,
|
| 1400 |
+
"eval_runtime": 225.2876,
|
| 1401 |
+
"eval_samples_per_second": 912.212,
|
| 1402 |
+
"eval_steps_per_second": 57.016,
|
| 1403 |
+
"step": 1016000
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 4.2,
|
| 1407 |
+
"learning_rate": 5.736679729842408e-06,
|
| 1408 |
+
"loss": 2.8085,
|
| 1409 |
+
"step": 1024000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 4.2,
|
| 1413 |
+
"eval_loss": 2.8393983840942383,
|
| 1414 |
+
"eval_runtime": 224.8219,
|
| 1415 |
+
"eval_samples_per_second": 914.101,
|
| 1416 |
+
"eval_steps_per_second": 57.134,
|
| 1417 |
+
"step": 1024000
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 4.23,
|
| 1421 |
+
"eval_loss": 2.835977077484131,
|
| 1422 |
+
"eval_runtime": 226.9491,
|
| 1423 |
+
"eval_samples_per_second": 905.533,
|
| 1424 |
+
"eval_steps_per_second": 56.599,
|
| 1425 |
+
"step": 1032000
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 4.26,
|
| 1429 |
+
"learning_rate": 5.669974151588427e-06,
|
| 1430 |
+
"loss": 2.8179,
|
| 1431 |
+
"step": 1040000
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 4.26,
|
| 1435 |
+
"eval_loss": 2.8393590450286865,
|
| 1436 |
+
"eval_runtime": 226.0543,
|
| 1437 |
+
"eval_samples_per_second": 909.118,
|
| 1438 |
+
"eval_steps_per_second": 56.823,
|
| 1439 |
+
"step": 1040000
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 4.29,
|
| 1443 |
+
"eval_loss": 2.840928077697754,
|
| 1444 |
+
"eval_runtime": 226.338,
|
| 1445 |
+
"eval_samples_per_second": 907.978,
|
| 1446 |
+
"eval_steps_per_second": 56.751,
|
| 1447 |
+
"step": 1048000
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 4.33,
|
| 1451 |
+
"learning_rate": 5.603268573334446e-06,
|
| 1452 |
+
"loss": 2.8212,
|
| 1453 |
+
"step": 1056000
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 4.33,
|
| 1457 |
+
"eval_loss": 2.834209442138672,
|
| 1458 |
+
"eval_runtime": 225.5106,
|
| 1459 |
+
"eval_samples_per_second": 911.31,
|
| 1460 |
+
"eval_steps_per_second": 56.96,
|
| 1461 |
+
"step": 1056000
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 4.36,
|
| 1465 |
+
"eval_loss": 2.8416330814361572,
|
| 1466 |
+
"eval_runtime": 226.856,
|
| 1467 |
+
"eval_samples_per_second": 905.905,
|
| 1468 |
+
"eval_steps_per_second": 56.622,
|
| 1469 |
+
"step": 1064000
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 4.39,
|
| 1473 |
+
"learning_rate": 5.536562995080464e-06,
|
| 1474 |
+
"loss": 2.8236,
|
| 1475 |
+
"step": 1072000
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 4.39,
|
| 1479 |
+
"eval_loss": 2.839299440383911,
|
| 1480 |
+
"eval_runtime": 228.289,
|
| 1481 |
+
"eval_samples_per_second": 900.218,
|
| 1482 |
+
"eval_steps_per_second": 56.266,
|
| 1483 |
+
"step": 1072000
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 4.43,
|
| 1487 |
+
"eval_loss": 2.8458597660064697,
|
| 1488 |
+
"eval_runtime": 228.7211,
|
| 1489 |
+
"eval_samples_per_second": 898.518,
|
| 1490 |
+
"eval_steps_per_second": 56.16,
|
| 1491 |
+
"step": 1080000
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 4.46,
|
| 1495 |
+
"learning_rate": 5.469857416826483e-06,
|
| 1496 |
+
"loss": 2.8134,
|
| 1497 |
+
"step": 1088000
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 4.46,
|
| 1501 |
+
"eval_loss": 2.844362497329712,
|
| 1502 |
+
"eval_runtime": 227.78,
|
| 1503 |
+
"eval_samples_per_second": 902.23,
|
| 1504 |
+
"eval_steps_per_second": 56.392,
|
| 1505 |
+
"step": 1088000
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 4.49,
|
| 1509 |
+
"eval_loss": 2.843454360961914,
|
| 1510 |
+
"eval_runtime": 228.7834,
|
| 1511 |
+
"eval_samples_per_second": 898.273,
|
| 1512 |
+
"eval_steps_per_second": 56.145,
|
| 1513 |
+
"step": 1096000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 4.52,
|
| 1517 |
+
"learning_rate": 5.403151838572501e-06,
|
| 1518 |
+
"loss": 2.8196,
|
| 1519 |
+
"step": 1104000
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 4.52,
|
| 1523 |
+
"eval_loss": 2.848449945449829,
|
| 1524 |
+
"eval_runtime": 228.2467,
|
| 1525 |
+
"eval_samples_per_second": 900.385,
|
| 1526 |
+
"eval_steps_per_second": 56.277,
|
| 1527 |
+
"step": 1104000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 4.56,
|
| 1531 |
+
"eval_loss": 2.8537204265594482,
|
| 1532 |
+
"eval_runtime": 231.8913,
|
| 1533 |
+
"eval_samples_per_second": 886.234,
|
| 1534 |
+
"eval_steps_per_second": 55.392,
|
| 1535 |
+
"step": 1112000
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 4.59,
|
| 1539 |
+
"learning_rate": 5.33644626031852e-06,
|
| 1540 |
+
"loss": 2.8284,
|
| 1541 |
+
"step": 1120000
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 4.59,
|
| 1545 |
+
"eval_loss": 2.8540537357330322,
|
| 1546 |
+
"eval_runtime": 230.1324,
|
| 1547 |
+
"eval_samples_per_second": 893.008,
|
| 1548 |
+
"eval_steps_per_second": 55.816,
|
| 1549 |
+
"step": 1120000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 4.62,
|
| 1553 |
+
"eval_loss": 2.8400697708129883,
|
| 1554 |
+
"eval_runtime": 230.0912,
|
| 1555 |
+
"eval_samples_per_second": 893.167,
|
| 1556 |
+
"eval_steps_per_second": 55.826,
|
| 1557 |
+
"step": 1128000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 4.65,
|
| 1561 |
+
"learning_rate": 5.269740682064538e-06,
|
| 1562 |
+
"loss": 2.8349,
|
| 1563 |
+
"step": 1136000
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 4.65,
|
| 1567 |
+
"eval_loss": 2.847630262374878,
|
| 1568 |
+
"eval_runtime": 227.545,
|
| 1569 |
+
"eval_samples_per_second": 903.162,
|
| 1570 |
+
"eval_steps_per_second": 56.45,
|
| 1571 |
+
"step": 1136000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 4.69,
|
| 1575 |
+
"eval_loss": 2.8476176261901855,
|
| 1576 |
+
"eval_runtime": 227.3305,
|
| 1577 |
+
"eval_samples_per_second": 904.014,
|
| 1578 |
+
"eval_steps_per_second": 56.504,
|
| 1579 |
+
"step": 1144000
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 4.72,
|
| 1583 |
+
"learning_rate": 5.203035103810556e-06,
|
| 1584 |
+
"loss": 2.8171,
|
| 1585 |
+
"step": 1152000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 4.72,
|
| 1589 |
+
"eval_loss": 2.843763589859009,
|
| 1590 |
+
"eval_runtime": 232.1395,
|
| 1591 |
+
"eval_samples_per_second": 885.287,
|
| 1592 |
+
"eval_steps_per_second": 55.333,
|
| 1593 |
+
"step": 1152000
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 4.75,
|
| 1597 |
+
"eval_loss": 2.853471040725708,
|
| 1598 |
+
"eval_runtime": 229.6265,
|
| 1599 |
+
"eval_samples_per_second": 894.975,
|
| 1600 |
+
"eval_steps_per_second": 55.939,
|
| 1601 |
+
"step": 1160000
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 4.79,
|
| 1605 |
+
"learning_rate": 5.136329525556575e-06,
|
| 1606 |
+
"loss": 2.8264,
|
| 1607 |
+
"step": 1168000
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 4.79,
|
| 1611 |
+
"eval_loss": 2.842829704284668,
|
| 1612 |
+
"eval_runtime": 228.2848,
|
| 1613 |
+
"eval_samples_per_second": 900.235,
|
| 1614 |
+
"eval_steps_per_second": 56.267,
|
| 1615 |
+
"step": 1168000
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 4.82,
|
| 1619 |
+
"eval_loss": 2.855221748352051,
|
| 1620 |
+
"eval_runtime": 229.9347,
|
| 1621 |
+
"eval_samples_per_second": 893.775,
|
| 1622 |
+
"eval_steps_per_second": 55.864,
|
| 1623 |
+
"step": 1176000
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 4.85,
|
| 1627 |
+
"learning_rate": 5.0696239473025935e-06,
|
| 1628 |
+
"loss": 2.8335,
|
| 1629 |
+
"step": 1184000
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 4.85,
|
| 1633 |
+
"eval_loss": 2.857285261154175,
|
| 1634 |
+
"eval_runtime": 229.2467,
|
| 1635 |
+
"eval_samples_per_second": 896.458,
|
| 1636 |
+
"eval_steps_per_second": 56.031,
|
| 1637 |
+
"step": 1184000
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 4.88,
|
| 1641 |
+
"eval_loss": 2.850496292114258,
|
| 1642 |
+
"eval_runtime": 228.4453,
|
| 1643 |
+
"eval_samples_per_second": 899.603,
|
| 1644 |
+
"eval_steps_per_second": 56.228,
|
| 1645 |
+
"step": 1192000
|
| 1646 |
+
},
|
| 1647 |
+
{
|
| 1648 |
+
"epoch": 4.92,
|
| 1649 |
+
"learning_rate": 5.002918369048611e-06,
|
| 1650 |
+
"loss": 2.8351,
|
| 1651 |
+
"step": 1200000
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 4.92,
|
| 1655 |
+
"eval_loss": 2.8511581420898438,
|
| 1656 |
+
"eval_runtime": 228.7507,
|
| 1657 |
+
"eval_samples_per_second": 898.401,
|
| 1658 |
+
"eval_steps_per_second": 56.153,
|
| 1659 |
+
"step": 1200000
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 4.95,
|
| 1663 |
+
"eval_loss": 2.8499553203582764,
|
| 1664 |
+
"eval_runtime": 230.0891,
|
| 1665 |
+
"eval_samples_per_second": 893.176,
|
| 1666 |
+
"eval_steps_per_second": 55.826,
|
| 1667 |
+
"step": 1208000
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 4.98,
|
| 1671 |
+
"learning_rate": 4.936212790794631e-06,
|
| 1672 |
+
"loss": 2.8366,
|
| 1673 |
+
"step": 1216000
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 4.98,
|
| 1677 |
+
"eval_loss": 2.8569633960723877,
|
| 1678 |
+
"eval_runtime": 229.1877,
|
| 1679 |
+
"eval_samples_per_second": 896.689,
|
| 1680 |
+
"eval_steps_per_second": 56.046,
|
| 1681 |
+
"step": 1216000
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 5.02,
|
| 1685 |
+
"eval_loss": 2.846999168395996,
|
| 1686 |
+
"eval_runtime": 228.106,
|
| 1687 |
+
"eval_samples_per_second": 900.941,
|
| 1688 |
+
"eval_steps_per_second": 56.312,
|
| 1689 |
+
"step": 1224000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 5.05,
|
| 1693 |
+
"learning_rate": 4.869507212540649e-06,
|
| 1694 |
+
"loss": 2.8257,
|
| 1695 |
+
"step": 1232000
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 5.05,
|
| 1699 |
+
"eval_loss": 2.8638033866882324,
|
| 1700 |
+
"eval_runtime": 227.5474,
|
| 1701 |
+
"eval_samples_per_second": 903.153,
|
| 1702 |
+
"eval_steps_per_second": 56.45,
|
| 1703 |
+
"step": 1232000
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 5.08,
|
| 1707 |
+
"eval_loss": 2.8511650562286377,
|
| 1708 |
+
"eval_runtime": 227.331,
|
| 1709 |
+
"eval_samples_per_second": 904.012,
|
| 1710 |
+
"eval_steps_per_second": 56.504,
|
| 1711 |
+
"step": 1240000
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 5.11,
|
| 1715 |
+
"learning_rate": 4.802801634286667e-06,
|
| 1716 |
+
"loss": 2.8351,
|
| 1717 |
+
"step": 1248000
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 5.11,
|
| 1721 |
+
"eval_loss": 2.8640658855438232,
|
| 1722 |
+
"eval_runtime": 227.1737,
|
| 1723 |
+
"eval_samples_per_second": 904.638,
|
| 1724 |
+
"eval_steps_per_second": 56.543,
|
| 1725 |
+
"step": 1248000
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 5.15,
|
| 1729 |
+
"eval_loss": 2.8679511547088623,
|
| 1730 |
+
"eval_runtime": 227.7339,
|
| 1731 |
+
"eval_samples_per_second": 902.413,
|
| 1732 |
+
"eval_steps_per_second": 56.404,
|
| 1733 |
+
"step": 1256000
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 5.18,
|
| 1737 |
+
"learning_rate": 4.7360960560326865e-06,
|
| 1738 |
+
"loss": 2.8272,
|
| 1739 |
+
"step": 1264000
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 5.18,
|
| 1743 |
+
"eval_loss": 2.8521032333374023,
|
| 1744 |
+
"eval_runtime": 226.8252,
|
| 1745 |
+
"eval_samples_per_second": 906.028,
|
| 1746 |
+
"eval_steps_per_second": 56.63,
|
| 1747 |
+
"step": 1264000
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 5.21,
|
| 1751 |
+
"eval_loss": 2.8616464138031006,
|
| 1752 |
+
"eval_runtime": 226.0986,
|
| 1753 |
+
"eval_samples_per_second": 908.94,
|
| 1754 |
+
"eval_steps_per_second": 56.812,
|
| 1755 |
+
"step": 1272000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 5.24,
|
| 1759 |
+
"learning_rate": 4.669390477778704e-06,
|
| 1760 |
+
"loss": 2.8346,
|
| 1761 |
+
"step": 1280000
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 5.24,
|
| 1765 |
+
"eval_loss": 2.8545360565185547,
|
| 1766 |
+
"eval_runtime": 226.6533,
|
| 1767 |
+
"eval_samples_per_second": 906.715,
|
| 1768 |
+
"eval_steps_per_second": 56.672,
|
| 1769 |
+
"step": 1280000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 5.28,
|
| 1773 |
+
"eval_loss": 2.847703695297241,
|
| 1774 |
+
"eval_runtime": 226.3926,
|
| 1775 |
+
"eval_samples_per_second": 907.759,
|
| 1776 |
+
"eval_steps_per_second": 56.738,
|
| 1777 |
+
"step": 1288000
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 5.31,
|
| 1781 |
+
"learning_rate": 4.602684899524723e-06,
|
| 1782 |
+
"loss": 2.8211,
|
| 1783 |
+
"step": 1296000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 5.31,
|
| 1787 |
+
"eval_loss": 2.860240936279297,
|
| 1788 |
+
"eval_runtime": 229.1022,
|
| 1789 |
+
"eval_samples_per_second": 897.023,
|
| 1790 |
+
"eval_steps_per_second": 56.067,
|
| 1791 |
+
"step": 1296000
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 5.34,
|
| 1795 |
+
"eval_loss": 2.857428550720215,
|
| 1796 |
+
"eval_runtime": 227.3841,
|
| 1797 |
+
"eval_samples_per_second": 903.801,
|
| 1798 |
+
"eval_steps_per_second": 56.49,
|
| 1799 |
+
"step": 1304000
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 5.38,
|
| 1803 |
+
"learning_rate": 4.5359793212707415e-06,
|
| 1804 |
+
"loss": 2.8302,
|
| 1805 |
+
"step": 1312000
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 5.38,
|
| 1809 |
+
"eval_loss": 2.8489556312561035,
|
| 1810 |
+
"eval_runtime": 227.6538,
|
| 1811 |
+
"eval_samples_per_second": 902.73,
|
| 1812 |
+
"eval_steps_per_second": 56.423,
|
| 1813 |
+
"step": 1312000
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 5.41,
|
| 1817 |
+
"eval_loss": 2.8547353744506836,
|
| 1818 |
+
"eval_runtime": 226.9648,
|
| 1819 |
+
"eval_samples_per_second": 905.471,
|
| 1820 |
+
"eval_steps_per_second": 56.595,
|
| 1821 |
+
"step": 1320000
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 5.44,
|
| 1825 |
+
"learning_rate": 4.46927374301676e-06,
|
| 1826 |
+
"loss": 2.8317,
|
| 1827 |
+
"step": 1328000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 5.44,
|
| 1831 |
+
"eval_loss": 2.8535571098327637,
|
| 1832 |
+
"eval_runtime": 228.1487,
|
| 1833 |
+
"eval_samples_per_second": 900.772,
|
| 1834 |
+
"eval_steps_per_second": 56.301,
|
| 1835 |
+
"step": 1328000
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 5.47,
|
| 1839 |
+
"eval_loss": 2.855344295501709,
|
| 1840 |
+
"eval_runtime": 227.3885,
|
| 1841 |
+
"eval_samples_per_second": 903.783,
|
| 1842 |
+
"eval_steps_per_second": 56.489,
|
| 1843 |
+
"step": 1336000
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 5.51,
|
| 1847 |
+
"learning_rate": 4.402568164762779e-06,
|
| 1848 |
+
"loss": 2.83,
|
| 1849 |
+
"step": 1344000
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 5.51,
|
| 1853 |
+
"eval_loss": 2.853558301925659,
|
| 1854 |
+
"eval_runtime": 226.8054,
|
| 1855 |
+
"eval_samples_per_second": 906.107,
|
| 1856 |
+
"eval_steps_per_second": 56.634,
|
| 1857 |
+
"step": 1344000
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 5.54,
|
| 1861 |
+
"eval_loss": 2.8564646244049072,
|
| 1862 |
+
"eval_runtime": 227.7928,
|
| 1863 |
+
"eval_samples_per_second": 902.18,
|
| 1864 |
+
"eval_steps_per_second": 56.389,
|
| 1865 |
+
"step": 1352000
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 5.57,
|
| 1869 |
+
"learning_rate": 4.335862586508797e-06,
|
| 1870 |
+
"loss": 2.8347,
|
| 1871 |
+
"step": 1360000
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 5.57,
|
| 1875 |
+
"eval_loss": 2.8444929122924805,
|
| 1876 |
+
"eval_runtime": 226.9232,
|
| 1877 |
+
"eval_samples_per_second": 905.637,
|
| 1878 |
+
"eval_steps_per_second": 56.605,
|
| 1879 |
+
"step": 1360000
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 5.61,
|
| 1883 |
+
"eval_loss": 2.854001998901367,
|
| 1884 |
+
"eval_runtime": 227.1661,
|
| 1885 |
+
"eval_samples_per_second": 904.668,
|
| 1886 |
+
"eval_steps_per_second": 56.545,
|
| 1887 |
+
"step": 1368000
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 5.64,
|
| 1891 |
+
"learning_rate": 4.269157008254816e-06,
|
| 1892 |
+
"loss": 2.8253,
|
| 1893 |
+
"step": 1376000
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 5.64,
|
| 1897 |
+
"eval_loss": 2.8630056381225586,
|
| 1898 |
+
"eval_runtime": 228.1691,
|
| 1899 |
+
"eval_samples_per_second": 900.692,
|
| 1900 |
+
"eval_steps_per_second": 56.296,
|
| 1901 |
+
"step": 1376000
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 5.67,
|
| 1905 |
+
"eval_loss": 2.8591771125793457,
|
| 1906 |
+
"eval_runtime": 228.3462,
|
| 1907 |
+
"eval_samples_per_second": 899.993,
|
| 1908 |
+
"eval_steps_per_second": 56.252,
|
| 1909 |
+
"step": 1384000
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 5.7,
|
| 1913 |
+
"learning_rate": 4.202451430000834e-06,
|
| 1914 |
+
"loss": 2.8237,
|
| 1915 |
+
"step": 1392000
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 5.7,
|
| 1919 |
+
"eval_loss": 2.8634839057922363,
|
| 1920 |
+
"eval_runtime": 227.2681,
|
| 1921 |
+
"eval_samples_per_second": 904.262,
|
| 1922 |
+
"eval_steps_per_second": 56.519,
|
| 1923 |
+
"step": 1392000
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 5.74,
|
| 1927 |
+
"eval_loss": 2.862060308456421,
|
| 1928 |
+
"eval_runtime": 226.8521,
|
| 1929 |
+
"eval_samples_per_second": 905.921,
|
| 1930 |
+
"eval_steps_per_second": 56.623,
|
| 1931 |
+
"step": 1400000
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 5.77,
|
| 1935 |
+
"learning_rate": 4.135745851746852e-06,
|
| 1936 |
+
"loss": 2.8364,
|
| 1937 |
+
"step": 1408000
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 5.77,
|
| 1941 |
+
"eval_loss": 2.854534864425659,
|
| 1942 |
+
"eval_runtime": 228.0962,
|
| 1943 |
+
"eval_samples_per_second": 900.98,
|
| 1944 |
+
"eval_steps_per_second": 56.314,
|
| 1945 |
+
"step": 1408000
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 5.8,
|
| 1949 |
+
"eval_loss": 2.8681845664978027,
|
| 1950 |
+
"eval_runtime": 228.9004,
|
| 1951 |
+
"eval_samples_per_second": 897.814,
|
| 1952 |
+
"eval_steps_per_second": 56.116,
|
| 1953 |
+
"step": 1416000
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 5.84,
|
| 1957 |
+
"learning_rate": 4.069040273492872e-06,
|
| 1958 |
+
"loss": 2.8289,
|
| 1959 |
+
"step": 1424000
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 5.84,
|
| 1963 |
+
"eval_loss": 2.867513418197632,
|
| 1964 |
+
"eval_runtime": 228.9233,
|
| 1965 |
+
"eval_samples_per_second": 897.724,
|
| 1966 |
+
"eval_steps_per_second": 56.111,
|
| 1967 |
+
"step": 1424000
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 5.87,
|
| 1971 |
+
"eval_loss": 2.859713077545166,
|
| 1972 |
+
"eval_runtime": 228.0494,
|
| 1973 |
+
"eval_samples_per_second": 901.164,
|
| 1974 |
+
"eval_steps_per_second": 56.326,
|
| 1975 |
+
"step": 1432000
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 5.9,
|
| 1979 |
+
"learning_rate": 4.0023346952388895e-06,
|
| 1980 |
+
"loss": 2.8327,
|
| 1981 |
+
"step": 1440000
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 5.9,
|
| 1985 |
+
"eval_loss": 2.8727521896362305,
|
| 1986 |
+
"eval_runtime": 228.1742,
|
| 1987 |
+
"eval_samples_per_second": 900.672,
|
| 1988 |
+
"eval_steps_per_second": 56.295,
|
| 1989 |
+
"step": 1440000
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 5.93,
|
| 1993 |
+
"eval_loss": 2.8643882274627686,
|
| 1994 |
+
"eval_runtime": 228.4395,
|
| 1995 |
+
"eval_samples_per_second": 899.626,
|
| 1996 |
+
"eval_steps_per_second": 56.229,
|
| 1997 |
+
"step": 1448000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 5.97,
|
| 2001 |
+
"learning_rate": 3.935629116984908e-06,
|
| 2002 |
+
"loss": 2.8407,
|
| 2003 |
+
"step": 1456000
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 5.97,
|
| 2007 |
+
"eval_loss": 2.864025354385376,
|
| 2008 |
+
"eval_runtime": 228.8691,
|
| 2009 |
+
"eval_samples_per_second": 897.937,
|
| 2010 |
+
"eval_steps_per_second": 56.124,
|
| 2011 |
+
"step": 1456000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 6.0,
|
| 2015 |
+
"eval_loss": 2.8669943809509277,
|
| 2016 |
+
"eval_runtime": 227.4377,
|
| 2017 |
+
"eval_samples_per_second": 903.588,
|
| 2018 |
+
"eval_steps_per_second": 56.477,
|
| 2019 |
+
"step": 1464000
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 6.03,
|
| 2023 |
+
"learning_rate": 3.868923538730927e-06,
|
| 2024 |
+
"loss": 2.8349,
|
| 2025 |
+
"step": 1472000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 6.03,
|
| 2029 |
+
"eval_loss": 2.8555285930633545,
|
| 2030 |
+
"eval_runtime": 228.413,
|
| 2031 |
+
"eval_samples_per_second": 899.73,
|
| 2032 |
+
"eval_steps_per_second": 56.236,
|
| 2033 |
+
"step": 1472000
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 6.06,
|
| 2037 |
+
"eval_loss": 2.8778481483459473,
|
| 2038 |
+
"eval_runtime": 228.1335,
|
| 2039 |
+
"eval_samples_per_second": 900.832,
|
| 2040 |
+
"eval_steps_per_second": 56.305,
|
| 2041 |
+
"step": 1480000
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 6.1,
|
| 2045 |
+
"learning_rate": 3.8022179604769453e-06,
|
| 2046 |
+
"loss": 2.8395,
|
| 2047 |
+
"step": 1488000
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 6.1,
|
| 2051 |
+
"eval_loss": 2.8752517700195312,
|
| 2052 |
+
"eval_runtime": 229.2602,
|
| 2053 |
+
"eval_samples_per_second": 896.405,
|
| 2054 |
+
"eval_steps_per_second": 56.028,
|
| 2055 |
+
"step": 1488000
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 6.13,
|
| 2059 |
+
"eval_loss": 2.865680694580078,
|
| 2060 |
+
"eval_runtime": 228.1396,
|
| 2061 |
+
"eval_samples_per_second": 900.808,
|
| 2062 |
+
"eval_steps_per_second": 56.303,
|
| 2063 |
+
"step": 1496000
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 6.16,
|
| 2067 |
+
"learning_rate": 3.735512382222964e-06,
|
| 2068 |
+
"loss": 2.8364,
|
| 2069 |
+
"step": 1504000
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 6.16,
|
| 2073 |
+
"eval_loss": 2.864389419555664,
|
| 2074 |
+
"eval_runtime": 228.8163,
|
| 2075 |
+
"eval_samples_per_second": 898.144,
|
| 2076 |
+
"eval_steps_per_second": 56.137,
|
| 2077 |
+
"step": 1504000
|
| 2078 |
+
},
|
| 2079 |
+
{
|
| 2080 |
+
"epoch": 6.2,
|
| 2081 |
+
"eval_loss": 2.8669490814208984,
|
| 2082 |
+
"eval_runtime": 230.4793,
|
| 2083 |
+
"eval_samples_per_second": 891.663,
|
| 2084 |
+
"eval_steps_per_second": 55.732,
|
| 2085 |
+
"step": 1512000
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 6.23,
|
| 2089 |
+
"learning_rate": 3.668806803968982e-06,
|
| 2090 |
+
"loss": 2.85,
|
| 2091 |
+
"step": 1520000
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 6.23,
|
| 2095 |
+
"eval_loss": 2.863649845123291,
|
| 2096 |
+
"eval_runtime": 228.5034,
|
| 2097 |
+
"eval_samples_per_second": 899.374,
|
| 2098 |
+
"eval_steps_per_second": 56.214,
|
| 2099 |
+
"step": 1520000
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 6.26,
|
| 2103 |
+
"eval_loss": 2.868006944656372,
|
| 2104 |
+
"eval_runtime": 228.7374,
|
| 2105 |
+
"eval_samples_per_second": 898.454,
|
| 2106 |
+
"eval_steps_per_second": 56.156,
|
| 2107 |
+
"step": 1528000
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 6.29,
|
| 2111 |
+
"learning_rate": 3.6021012257150007e-06,
|
| 2112 |
+
"loss": 2.8359,
|
| 2113 |
+
"step": 1536000
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 6.29,
|
| 2117 |
+
"eval_loss": 2.8751797676086426,
|
| 2118 |
+
"eval_runtime": 228.8775,
|
| 2119 |
+
"eval_samples_per_second": 897.904,
|
| 2120 |
+
"eval_steps_per_second": 56.122,
|
| 2121 |
+
"step": 1536000
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 6.33,
|
| 2125 |
+
"eval_loss": 2.871007204055786,
|
| 2126 |
+
"eval_runtime": 229.7636,
|
| 2127 |
+
"eval_samples_per_second": 894.441,
|
| 2128 |
+
"eval_steps_per_second": 55.905,
|
| 2129 |
+
"step": 1544000
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 6.36,
|
| 2133 |
+
"learning_rate": 3.535395647461019e-06,
|
| 2134 |
+
"loss": 2.8451,
|
| 2135 |
+
"step": 1552000
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 6.36,
|
| 2139 |
+
"eval_loss": 2.876699686050415,
|
| 2140 |
+
"eval_runtime": 229.2961,
|
| 2141 |
+
"eval_samples_per_second": 896.265,
|
| 2142 |
+
"eval_steps_per_second": 56.019,
|
| 2143 |
+
"step": 1552000
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 6.39,
|
| 2147 |
+
"eval_loss": 2.882432699203491,
|
| 2148 |
+
"eval_runtime": 228.5189,
|
| 2149 |
+
"eval_samples_per_second": 899.313,
|
| 2150 |
+
"eval_steps_per_second": 56.21,
|
| 2151 |
+
"step": 1560000
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 6.43,
|
| 2155 |
+
"learning_rate": 3.468690069207038e-06,
|
| 2156 |
+
"loss": 2.8359,
|
| 2157 |
+
"step": 1568000
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 6.43,
|
| 2161 |
+
"eval_loss": 2.8723318576812744,
|
| 2162 |
+
"eval_runtime": 229.3117,
|
| 2163 |
+
"eval_samples_per_second": 896.204,
|
| 2164 |
+
"eval_steps_per_second": 56.015,
|
| 2165 |
+
"step": 1568000
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 6.46,
|
| 2169 |
+
"eval_loss": 2.87728214263916,
|
| 2170 |
+
"eval_runtime": 230.0667,
|
| 2171 |
+
"eval_samples_per_second": 893.263,
|
| 2172 |
+
"eval_steps_per_second": 55.832,
|
| 2173 |
+
"step": 1576000
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 6.49,
|
| 2177 |
+
"learning_rate": 3.4019844909530565e-06,
|
| 2178 |
+
"loss": 2.8546,
|
| 2179 |
+
"step": 1584000
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 6.49,
|
| 2183 |
+
"eval_loss": 2.875946044921875,
|
| 2184 |
+
"eval_runtime": 228.6285,
|
| 2185 |
+
"eval_samples_per_second": 898.882,
|
| 2186 |
+
"eval_steps_per_second": 56.183,
|
| 2187 |
+
"step": 1584000
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 6.52,
|
| 2191 |
+
"eval_loss": 2.8732051849365234,
|
| 2192 |
+
"eval_runtime": 231.6471,
|
| 2193 |
+
"eval_samples_per_second": 887.168,
|
| 2194 |
+
"eval_steps_per_second": 55.451,
|
| 2195 |
+
"step": 1592000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 6.56,
|
| 2199 |
+
"learning_rate": 3.3352789126990747e-06,
|
| 2200 |
+
"loss": 2.8395,
|
| 2201 |
+
"step": 1600000
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 6.56,
|
| 2205 |
+
"eval_loss": 2.8803162574768066,
|
| 2206 |
+
"eval_runtime": 230.2291,
|
| 2207 |
+
"eval_samples_per_second": 892.632,
|
| 2208 |
+
"eval_steps_per_second": 55.792,
|
| 2209 |
+
"step": 1600000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 6.59,
|
| 2213 |
+
"eval_loss": 2.8761353492736816,
|
| 2214 |
+
"eval_runtime": 229.013,
|
| 2215 |
+
"eval_samples_per_second": 897.373,
|
| 2216 |
+
"eval_steps_per_second": 56.089,
|
| 2217 |
+
"step": 1608000
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 6.62,
|
| 2221 |
+
"learning_rate": 3.2685733344450933e-06,
|
| 2222 |
+
"loss": 2.847,
|
| 2223 |
+
"step": 1616000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 6.62,
|
| 2227 |
+
"eval_loss": 2.880068302154541,
|
| 2228 |
+
"eval_runtime": 229.0344,
|
| 2229 |
+
"eval_samples_per_second": 897.289,
|
| 2230 |
+
"eval_steps_per_second": 56.083,
|
| 2231 |
+
"step": 1616000
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 6.65,
|
| 2235 |
+
"eval_loss": 2.873690366744995,
|
| 2236 |
+
"eval_runtime": 228.9162,
|
| 2237 |
+
"eval_samples_per_second": 897.752,
|
| 2238 |
+
"eval_steps_per_second": 56.112,
|
| 2239 |
+
"step": 1624000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 6.69,
|
| 2243 |
+
"learning_rate": 3.2018677561911115e-06,
|
| 2244 |
+
"loss": 2.8555,
|
| 2245 |
+
"step": 1632000
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 6.69,
|
| 2249 |
+
"eval_loss": 2.8797101974487305,
|
| 2250 |
+
"eval_runtime": 228.8191,
|
| 2251 |
+
"eval_samples_per_second": 898.133,
|
| 2252 |
+
"eval_steps_per_second": 56.136,
|
| 2253 |
+
"step": 1632000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 6.72,
|
| 2257 |
+
"eval_loss": 2.8781728744506836,
|
| 2258 |
+
"eval_runtime": 229.3564,
|
| 2259 |
+
"eval_samples_per_second": 896.029,
|
| 2260 |
+
"eval_steps_per_second": 56.005,
|
| 2261 |
+
"step": 1640000
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 6.75,
|
| 2265 |
+
"learning_rate": 3.1351621779371306e-06,
|
| 2266 |
+
"loss": 2.8377,
|
| 2267 |
+
"step": 1648000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 6.75,
|
| 2271 |
+
"eval_loss": 2.8825597763061523,
|
| 2272 |
+
"eval_runtime": 229.3988,
|
| 2273 |
+
"eval_samples_per_second": 895.863,
|
| 2274 |
+
"eval_steps_per_second": 55.994,
|
| 2275 |
+
"step": 1648000
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 6.79,
|
| 2279 |
+
"eval_loss": 2.879844903945923,
|
| 2280 |
+
"eval_runtime": 230.5279,
|
| 2281 |
+
"eval_samples_per_second": 891.476,
|
| 2282 |
+
"eval_steps_per_second": 55.72,
|
| 2283 |
+
"step": 1656000
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 6.82,
|
| 2287 |
+
"learning_rate": 3.0684565996831487e-06,
|
| 2288 |
+
"loss": 2.8517,
|
| 2289 |
+
"step": 1664000
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 6.82,
|
| 2293 |
+
"eval_loss": 2.8799245357513428,
|
| 2294 |
+
"eval_runtime": 230.1674,
|
| 2295 |
+
"eval_samples_per_second": 892.872,
|
| 2296 |
+
"eval_steps_per_second": 55.807,
|
| 2297 |
+
"step": 1664000
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 6.85,
|
| 2301 |
+
"eval_loss": 2.8835196495056152,
|
| 2302 |
+
"eval_runtime": 229.3852,
|
| 2303 |
+
"eval_samples_per_second": 895.917,
|
| 2304 |
+
"eval_steps_per_second": 55.998,
|
| 2305 |
+
"step": 1672000
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 6.88,
|
| 2309 |
+
"learning_rate": 3.0017510214291673e-06,
|
| 2310 |
+
"loss": 2.8526,
|
| 2311 |
+
"step": 1680000
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 6.88,
|
| 2315 |
+
"eval_loss": 2.8874735832214355,
|
| 2316 |
+
"eval_runtime": 229.3731,
|
| 2317 |
+
"eval_samples_per_second": 895.964,
|
| 2318 |
+
"eval_steps_per_second": 56.0,
|
| 2319 |
+
"step": 1680000
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 6.92,
|
| 2323 |
+
"eval_loss": 2.882905960083008,
|
| 2324 |
+
"eval_runtime": 231.0041,
|
| 2325 |
+
"eval_samples_per_second": 889.638,
|
| 2326 |
+
"eval_steps_per_second": 55.605,
|
| 2327 |
+
"step": 1688000
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 6.95,
|
| 2331 |
+
"learning_rate": 2.9350454431751855e-06,
|
| 2332 |
+
"loss": 2.8511,
|
| 2333 |
+
"step": 1696000
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 6.95,
|
| 2337 |
+
"eval_loss": 2.8908419609069824,
|
| 2338 |
+
"eval_runtime": 229.2389,
|
| 2339 |
+
"eval_samples_per_second": 896.488,
|
| 2340 |
+
"eval_steps_per_second": 56.033,
|
| 2341 |
+
"step": 1696000
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 6.98,
|
| 2345 |
+
"eval_loss": 2.8755762577056885,
|
| 2346 |
+
"eval_runtime": 230.9816,
|
| 2347 |
+
"eval_samples_per_second": 889.724,
|
| 2348 |
+
"eval_steps_per_second": 55.61,
|
| 2349 |
+
"step": 1704000
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 7.02,
|
| 2353 |
+
"learning_rate": 2.868339864921204e-06,
|
| 2354 |
+
"loss": 2.8606,
|
| 2355 |
+
"step": 1712000
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 7.02,
|
| 2359 |
+
"eval_loss": 2.882680654525757,
|
| 2360 |
+
"eval_runtime": 230.7858,
|
| 2361 |
+
"eval_samples_per_second": 890.479,
|
| 2362 |
+
"eval_steps_per_second": 55.658,
|
| 2363 |
+
"step": 1712000
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 7.05,
|
| 2367 |
+
"eval_loss": 2.884394407272339,
|
| 2368 |
+
"eval_runtime": 230.0457,
|
| 2369 |
+
"eval_samples_per_second": 893.344,
|
| 2370 |
+
"eval_steps_per_second": 55.837,
|
| 2371 |
+
"step": 1720000
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 7.08,
|
| 2375 |
+
"learning_rate": 2.801634286667223e-06,
|
| 2376 |
+
"loss": 2.852,
|
| 2377 |
+
"step": 1728000
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 7.08,
|
| 2381 |
+
"eval_loss": 2.8865275382995605,
|
| 2382 |
+
"eval_runtime": 229.7673,
|
| 2383 |
+
"eval_samples_per_second": 894.427,
|
| 2384 |
+
"eval_steps_per_second": 55.904,
|
| 2385 |
+
"step": 1728000
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 7.11,
|
| 2389 |
+
"eval_loss": 2.8909904956817627,
|
| 2390 |
+
"eval_runtime": 230.4684,
|
| 2391 |
+
"eval_samples_per_second": 891.706,
|
| 2392 |
+
"eval_steps_per_second": 55.734,
|
| 2393 |
+
"step": 1736000
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 7.15,
|
| 2397 |
+
"learning_rate": 2.7349287084132413e-06,
|
| 2398 |
+
"loss": 2.8319,
|
| 2399 |
+
"step": 1744000
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 7.15,
|
| 2403 |
+
"eval_loss": 2.884788751602173,
|
| 2404 |
+
"eval_runtime": 229.622,
|
| 2405 |
+
"eval_samples_per_second": 894.993,
|
| 2406 |
+
"eval_steps_per_second": 55.94,
|
| 2407 |
+
"step": 1744000
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 7.18,
|
| 2411 |
+
"eval_loss": 2.8916478157043457,
|
| 2412 |
+
"eval_runtime": 231.4898,
|
| 2413 |
+
"eval_samples_per_second": 887.771,
|
| 2414 |
+
"eval_steps_per_second": 55.488,
|
| 2415 |
+
"step": 1752000
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 7.21,
|
| 2419 |
+
"learning_rate": 2.66822313015926e-06,
|
| 2420 |
+
"loss": 2.842,
|
| 2421 |
+
"step": 1760000
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 7.21,
|
| 2425 |
+
"eval_loss": 2.8830392360687256,
|
| 2426 |
+
"eval_runtime": 230.1781,
|
| 2427 |
+
"eval_samples_per_second": 892.83,
|
| 2428 |
+
"eval_steps_per_second": 55.805,
|
| 2429 |
+
"step": 1760000
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 7.24,
|
| 2433 |
+
"eval_loss": 2.8850438594818115,
|
| 2434 |
+
"eval_runtime": 230.145,
|
| 2435 |
+
"eval_samples_per_second": 892.959,
|
| 2436 |
+
"eval_steps_per_second": 55.813,
|
| 2437 |
+
"step": 1768000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 7.28,
|
| 2441 |
+
"learning_rate": 2.601517551905278e-06,
|
| 2442 |
+
"loss": 2.8421,
|
| 2443 |
+
"step": 1776000
|
| 2444 |
+
},
|
| 2445 |
+
{
|
| 2446 |
+
"epoch": 7.28,
|
| 2447 |
+
"eval_loss": 2.875260591506958,
|
| 2448 |
+
"eval_runtime": 230.4814,
|
| 2449 |
+
"eval_samples_per_second": 891.655,
|
| 2450 |
+
"eval_steps_per_second": 55.731,
|
| 2451 |
+
"step": 1776000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 7.31,
|
| 2455 |
+
"eval_loss": 2.8958091735839844,
|
| 2456 |
+
"eval_runtime": 230.6404,
|
| 2457 |
+
"eval_samples_per_second": 891.041,
|
| 2458 |
+
"eval_steps_per_second": 55.693,
|
| 2459 |
+
"step": 1784000
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 7.34,
|
| 2463 |
+
"learning_rate": 2.5348119736512967e-06,
|
| 2464 |
+
"loss": 2.8558,
|
| 2465 |
+
"step": 1792000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 7.34,
|
| 2469 |
+
"eval_loss": 2.8712708950042725,
|
| 2470 |
+
"eval_runtime": 229.8418,
|
| 2471 |
+
"eval_samples_per_second": 894.137,
|
| 2472 |
+
"eval_steps_per_second": 55.886,
|
| 2473 |
+
"step": 1792000
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 7.38,
|
| 2477 |
+
"eval_loss": 2.8743996620178223,
|
| 2478 |
+
"eval_runtime": 231.7487,
|
| 2479 |
+
"eval_samples_per_second": 886.779,
|
| 2480 |
+
"eval_steps_per_second": 55.426,
|
| 2481 |
+
"step": 1800000
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 7.41,
|
| 2485 |
+
"learning_rate": 2.4681063953973154e-06,
|
| 2486 |
+
"loss": 2.8382,
|
| 2487 |
+
"step": 1808000
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 7.41,
|
| 2491 |
+
"eval_loss": 2.890810966491699,
|
| 2492 |
+
"eval_runtime": 230.3071,
|
| 2493 |
+
"eval_samples_per_second": 892.33,
|
| 2494 |
+
"eval_steps_per_second": 55.773,
|
| 2495 |
+
"step": 1808000
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 7.44,
|
| 2499 |
+
"eval_loss": 2.8748860359191895,
|
| 2500 |
+
"eval_runtime": 230.3829,
|
| 2501 |
+
"eval_samples_per_second": 892.037,
|
| 2502 |
+
"eval_steps_per_second": 55.755,
|
| 2503 |
+
"step": 1816000
|
| 2504 |
+
},
|
| 2505 |
+
{
|
| 2506 |
+
"epoch": 7.47,
|
| 2507 |
+
"learning_rate": 2.4014008171433335e-06,
|
| 2508 |
+
"loss": 2.8508,
|
| 2509 |
+
"step": 1824000
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 7.47,
|
| 2513 |
+
"eval_loss": 2.878951072692871,
|
| 2514 |
+
"eval_runtime": 230.3626,
|
| 2515 |
+
"eval_samples_per_second": 892.115,
|
| 2516 |
+
"eval_steps_per_second": 55.76,
|
| 2517 |
+
"step": 1824000
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 7.51,
|
| 2521 |
+
"eval_loss": 2.886600971221924,
|
| 2522 |
+
"eval_runtime": 230.5556,
|
| 2523 |
+
"eval_samples_per_second": 891.368,
|
| 2524 |
+
"eval_steps_per_second": 55.713,
|
| 2525 |
+
"step": 1832000
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 7.54,
|
| 2529 |
+
"learning_rate": 2.334695238889352e-06,
|
| 2530 |
+
"loss": 2.8477,
|
| 2531 |
+
"step": 1840000
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 7.54,
|
| 2535 |
+
"eval_loss": 2.8806040287017822,
|
| 2536 |
+
"eval_runtime": 230.3727,
|
| 2537 |
+
"eval_samples_per_second": 892.076,
|
| 2538 |
+
"eval_steps_per_second": 55.757,
|
| 2539 |
+
"step": 1840000
|
| 2540 |
+
},
|
| 2541 |
+
{
|
| 2542 |
+
"epoch": 7.57,
|
| 2543 |
+
"eval_loss": 2.8820905685424805,
|
| 2544 |
+
"eval_runtime": 230.4476,
|
| 2545 |
+
"eval_samples_per_second": 891.786,
|
| 2546 |
+
"eval_steps_per_second": 55.739,
|
| 2547 |
+
"step": 1848000
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 7.61,
|
| 2551 |
+
"learning_rate": 2.2679896606353707e-06,
|
| 2552 |
+
"loss": 2.8497,
|
| 2553 |
+
"step": 1856000
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 7.61,
|
| 2557 |
+
"eval_loss": 2.877049684524536,
|
| 2558 |
+
"eval_runtime": 230.7875,
|
| 2559 |
+
"eval_samples_per_second": 890.473,
|
| 2560 |
+
"eval_steps_per_second": 55.657,
|
| 2561 |
+
"step": 1856000
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 7.64,
|
| 2565 |
+
"eval_loss": 2.873206377029419,
|
| 2566 |
+
"eval_runtime": 230.6269,
|
| 2567 |
+
"eval_samples_per_second": 891.093,
|
| 2568 |
+
"eval_steps_per_second": 55.696,
|
| 2569 |
+
"step": 1864000
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 7.67,
|
| 2573 |
+
"learning_rate": 2.2012840823813894e-06,
|
| 2574 |
+
"loss": 2.8566,
|
| 2575 |
+
"step": 1872000
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 7.67,
|
| 2579 |
+
"eval_loss": 2.887915849685669,
|
| 2580 |
+
"eval_runtime": 230.6589,
|
| 2581 |
+
"eval_samples_per_second": 890.969,
|
| 2582 |
+
"eval_steps_per_second": 55.688,
|
| 2583 |
+
"step": 1872000
|
| 2584 |
+
},
|
| 2585 |
+
{
|
| 2586 |
+
"epoch": 7.7,
|
| 2587 |
+
"eval_loss": 2.876028537750244,
|
| 2588 |
+
"eval_runtime": 230.4372,
|
| 2589 |
+
"eval_samples_per_second": 891.826,
|
| 2590 |
+
"eval_steps_per_second": 55.742,
|
| 2591 |
+
"step": 1880000
|
| 2592 |
+
},
|
| 2593 |
+
{
|
| 2594 |
+
"epoch": 7.74,
|
| 2595 |
+
"learning_rate": 2.134578504127408e-06,
|
| 2596 |
+
"loss": 2.8527,
|
| 2597 |
+
"step": 1888000
|
| 2598 |
+
},
|
| 2599 |
+
{
|
| 2600 |
+
"epoch": 7.74,
|
| 2601 |
+
"eval_loss": 2.8763537406921387,
|
| 2602 |
+
"eval_runtime": 232.1508,
|
| 2603 |
+
"eval_samples_per_second": 885.244,
|
| 2604 |
+
"eval_steps_per_second": 55.33,
|
| 2605 |
+
"step": 1888000
|
| 2606 |
+
},
|
| 2607 |
+
{
|
| 2608 |
+
"epoch": 7.77,
|
| 2609 |
+
"eval_loss": 2.8837802410125732,
|
| 2610 |
+
"eval_runtime": 231.7421,
|
| 2611 |
+
"eval_samples_per_second": 886.805,
|
| 2612 |
+
"eval_steps_per_second": 55.428,
|
| 2613 |
+
"step": 1896000
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 7.8,
|
| 2617 |
+
"learning_rate": 2.067872925873426e-06,
|
| 2618 |
+
"loss": 2.8438,
|
| 2619 |
+
"step": 1904000
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 7.8,
|
| 2623 |
+
"eval_loss": 2.8954691886901855,
|
| 2624 |
+
"eval_runtime": 231.1122,
|
| 2625 |
+
"eval_samples_per_second": 889.222,
|
| 2626 |
+
"eval_steps_per_second": 55.579,
|
| 2627 |
+
"step": 1904000
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 7.83,
|
| 2631 |
+
"eval_loss": 2.88919997215271,
|
| 2632 |
+
"eval_runtime": 232.0892,
|
| 2633 |
+
"eval_samples_per_second": 885.478,
|
| 2634 |
+
"eval_steps_per_second": 55.345,
|
| 2635 |
+
"step": 1912000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 7.87,
|
| 2639 |
+
"learning_rate": 2.0011673476194448e-06,
|
| 2640 |
+
"loss": 2.8422,
|
| 2641 |
+
"step": 1920000
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 7.87,
|
| 2645 |
+
"eval_loss": 2.883701801300049,
|
| 2646 |
+
"eval_runtime": 231.4957,
|
| 2647 |
+
"eval_samples_per_second": 887.749,
|
| 2648 |
+
"eval_steps_per_second": 55.487,
|
| 2649 |
+
"step": 1920000
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 7.9,
|
| 2653 |
+
"eval_loss": 2.89695143699646,
|
| 2654 |
+
"eval_runtime": 230.9788,
|
| 2655 |
+
"eval_samples_per_second": 889.735,
|
| 2656 |
+
"eval_steps_per_second": 55.611,
|
| 2657 |
+
"step": 1928000
|
| 2658 |
+
},
|
| 2659 |
+
{
|
| 2660 |
+
"epoch": 7.93,
|
| 2661 |
+
"learning_rate": 1.9344617693654634e-06,
|
| 2662 |
+
"loss": 2.8521,
|
| 2663 |
+
"step": 1936000
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 7.93,
|
| 2667 |
+
"eval_loss": 2.8805034160614014,
|
| 2668 |
+
"eval_runtime": 231.7434,
|
| 2669 |
+
"eval_samples_per_second": 886.8,
|
| 2670 |
+
"eval_steps_per_second": 55.428,
|
| 2671 |
+
"step": 1936000
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 7.97,
|
| 2675 |
+
"eval_loss": 2.8818843364715576,
|
| 2676 |
+
"eval_runtime": 230.9357,
|
| 2677 |
+
"eval_samples_per_second": 889.901,
|
| 2678 |
+
"eval_steps_per_second": 55.622,
|
| 2679 |
+
"step": 1944000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 8.0,
|
| 2683 |
+
"learning_rate": 1.867756191111482e-06,
|
| 2684 |
+
"loss": 2.8562,
|
| 2685 |
+
"step": 1952000
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 8.0,
|
| 2689 |
+
"eval_loss": 2.877063751220703,
|
| 2690 |
+
"eval_runtime": 230.6656,
|
| 2691 |
+
"eval_samples_per_second": 890.943,
|
| 2692 |
+
"eval_steps_per_second": 55.687,
|
| 2693 |
+
"step": 1952000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 8.03,
|
| 2697 |
+
"eval_loss": 2.881901502609253,
|
| 2698 |
+
"eval_runtime": 231.9508,
|
| 2699 |
+
"eval_samples_per_second": 886.007,
|
| 2700 |
+
"eval_steps_per_second": 55.378,
|
| 2701 |
+
"step": 1960000
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 8.06,
|
| 2705 |
+
"learning_rate": 1.8010506128575004e-06,
|
| 2706 |
+
"loss": 2.8417,
|
| 2707 |
+
"step": 1968000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 8.06,
|
| 2711 |
+
"eval_loss": 2.883180618286133,
|
| 2712 |
+
"eval_runtime": 231.1061,
|
| 2713 |
+
"eval_samples_per_second": 889.245,
|
| 2714 |
+
"eval_steps_per_second": 55.581,
|
| 2715 |
+
"step": 1968000
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 8.1,
|
| 2719 |
+
"eval_loss": 2.8927578926086426,
|
| 2720 |
+
"eval_runtime": 232.292,
|
| 2721 |
+
"eval_samples_per_second": 884.705,
|
| 2722 |
+
"eval_steps_per_second": 55.297,
|
| 2723 |
+
"step": 1976000
|
| 2724 |
+
},
|
| 2725 |
+
{
|
| 2726 |
+
"epoch": 8.13,
|
| 2727 |
+
"learning_rate": 1.734345034603519e-06,
|
| 2728 |
+
"loss": 2.8493,
|
| 2729 |
+
"step": 1984000
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 8.13,
|
| 2733 |
+
"eval_loss": 2.8891210556030273,
|
| 2734 |
+
"eval_runtime": 232.2356,
|
| 2735 |
+
"eval_samples_per_second": 884.92,
|
| 2736 |
+
"eval_steps_per_second": 55.31,
|
| 2737 |
+
"step": 1984000
|
| 2738 |
+
},
|
| 2739 |
+
{
|
| 2740 |
+
"epoch": 8.16,
|
| 2741 |
+
"eval_loss": 2.8862810134887695,
|
| 2742 |
+
"eval_runtime": 233.5203,
|
| 2743 |
+
"eval_samples_per_second": 880.052,
|
| 2744 |
+
"eval_steps_per_second": 55.006,
|
| 2745 |
+
"step": 1992000
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 8.2,
|
| 2749 |
+
"learning_rate": 1.6676394563495374e-06,
|
| 2750 |
+
"loss": 2.8549,
|
| 2751 |
+
"step": 2000000
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 8.2,
|
| 2755 |
+
"eval_loss": 2.876460313796997,
|
| 2756 |
+
"eval_runtime": 232.4349,
|
| 2757 |
+
"eval_samples_per_second": 884.162,
|
| 2758 |
+
"eval_steps_per_second": 55.263,
|
| 2759 |
+
"step": 2000000
|
| 2760 |
+
},
|
| 2761 |
+
{
|
| 2762 |
+
"epoch": 8.23,
|
| 2763 |
+
"eval_loss": 2.892146110534668,
|
| 2764 |
+
"eval_runtime": 232.392,
|
| 2765 |
+
"eval_samples_per_second": 884.325,
|
| 2766 |
+
"eval_steps_per_second": 55.273,
|
| 2767 |
+
"step": 2008000
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 8.26,
|
| 2771 |
+
"learning_rate": 1.6009338780955558e-06,
|
| 2772 |
+
"loss": 2.8421,
|
| 2773 |
+
"step": 2016000
|
| 2774 |
+
},
|
| 2775 |
+
{
|
| 2776 |
+
"epoch": 8.26,
|
| 2777 |
+
"eval_loss": 2.897282600402832,
|
| 2778 |
+
"eval_runtime": 234.5662,
|
| 2779 |
+
"eval_samples_per_second": 876.128,
|
| 2780 |
+
"eval_steps_per_second": 54.761,
|
| 2781 |
+
"step": 2016000
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 8.29,
|
| 2785 |
+
"eval_loss": 2.884657621383667,
|
| 2786 |
+
"eval_runtime": 232.2059,
|
| 2787 |
+
"eval_samples_per_second": 885.033,
|
| 2788 |
+
"eval_steps_per_second": 55.317,
|
| 2789 |
+
"step": 2024000
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 8.33,
|
| 2793 |
+
"learning_rate": 1.5342282998415744e-06,
|
| 2794 |
+
"loss": 2.8451,
|
| 2795 |
+
"step": 2032000
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 8.33,
|
| 2799 |
+
"eval_loss": 2.885942220687866,
|
| 2800 |
+
"eval_runtime": 236.2429,
|
| 2801 |
+
"eval_samples_per_second": 869.91,
|
| 2802 |
+
"eval_steps_per_second": 54.372,
|
| 2803 |
+
"step": 2032000
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 8.36,
|
| 2807 |
+
"eval_loss": 2.8866655826568604,
|
| 2808 |
+
"eval_runtime": 233.4019,
|
| 2809 |
+
"eval_samples_per_second": 880.498,
|
| 2810 |
+
"eval_steps_per_second": 55.034,
|
| 2811 |
+
"step": 2040000
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 8.39,
|
| 2815 |
+
"learning_rate": 1.4675227215875928e-06,
|
| 2816 |
+
"loss": 2.8465,
|
| 2817 |
+
"step": 2048000
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 8.39,
|
| 2821 |
+
"eval_loss": 2.8853070735931396,
|
| 2822 |
+
"eval_runtime": 233.5685,
|
| 2823 |
+
"eval_samples_per_second": 879.87,
|
| 2824 |
+
"eval_steps_per_second": 54.995,
|
| 2825 |
+
"step": 2048000
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 8.42,
|
| 2829 |
+
"eval_loss": 2.885338306427002,
|
| 2830 |
+
"eval_runtime": 234.5164,
|
| 2831 |
+
"eval_samples_per_second": 876.314,
|
| 2832 |
+
"eval_steps_per_second": 54.772,
|
| 2833 |
+
"step": 2056000
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 8.46,
|
| 2837 |
+
"learning_rate": 1.4008171433336116e-06,
|
| 2838 |
+
"loss": 2.8516,
|
| 2839 |
+
"step": 2064000
|
| 2840 |
+
},
|
| 2841 |
+
{
|
| 2842 |
+
"epoch": 8.46,
|
| 2843 |
+
"eval_loss": 2.8797264099121094,
|
| 2844 |
+
"eval_runtime": 232.9203,
|
| 2845 |
+
"eval_samples_per_second": 882.319,
|
| 2846 |
+
"eval_steps_per_second": 55.148,
|
| 2847 |
+
"step": 2064000
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 8.49,
|
| 2851 |
+
"eval_loss": 2.882495880126953,
|
| 2852 |
+
"eval_runtime": 232.5052,
|
| 2853 |
+
"eval_samples_per_second": 883.894,
|
| 2854 |
+
"eval_steps_per_second": 55.246,
|
| 2855 |
+
"step": 2072000
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 8.52,
|
| 2859 |
+
"learning_rate": 1.33411156507963e-06,
|
| 2860 |
+
"loss": 2.8519,
|
| 2861 |
+
"step": 2080000
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 8.52,
|
| 2865 |
+
"eval_loss": 2.8863439559936523,
|
| 2866 |
+
"eval_runtime": 232.936,
|
| 2867 |
+
"eval_samples_per_second": 882.26,
|
| 2868 |
+
"eval_steps_per_second": 55.144,
|
| 2869 |
+
"step": 2080000
|
| 2870 |
+
},
|
| 2871 |
+
{
|
| 2872 |
+
"epoch": 8.56,
|
| 2873 |
+
"eval_loss": 2.8823108673095703,
|
| 2874 |
+
"eval_runtime": 234.3106,
|
| 2875 |
+
"eval_samples_per_second": 877.084,
|
| 2876 |
+
"eval_steps_per_second": 54.82,
|
| 2877 |
+
"step": 2088000
|
| 2878 |
+
},
|
| 2879 |
+
{
|
| 2880 |
+
"epoch": 8.59,
|
| 2881 |
+
"learning_rate": 1.2674059868256484e-06,
|
| 2882 |
+
"loss": 2.8454,
|
| 2883 |
+
"step": 2096000
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 8.59,
|
| 2887 |
+
"eval_loss": 2.887014865875244,
|
| 2888 |
+
"eval_runtime": 236.1124,
|
| 2889 |
+
"eval_samples_per_second": 870.391,
|
| 2890 |
+
"eval_steps_per_second": 54.402,
|
| 2891 |
+
"step": 2096000
|
| 2892 |
+
},
|
| 2893 |
+
{
|
| 2894 |
+
"epoch": 8.62,
|
| 2895 |
+
"eval_loss": 2.8897511959075928,
|
| 2896 |
+
"eval_runtime": 234.9051,
|
| 2897 |
+
"eval_samples_per_second": 874.864,
|
| 2898 |
+
"eval_steps_per_second": 54.682,
|
| 2899 |
+
"step": 2104000
|
| 2900 |
+
},
|
| 2901 |
+
{
|
| 2902 |
+
"epoch": 8.65,
|
| 2903 |
+
"learning_rate": 1.2007004085716668e-06,
|
| 2904 |
+
"loss": 2.8428,
|
| 2905 |
+
"step": 2112000
|
| 2906 |
+
},
|
| 2907 |
+
{
|
| 2908 |
+
"epoch": 8.65,
|
| 2909 |
+
"eval_loss": 2.8753533363342285,
|
| 2910 |
+
"eval_runtime": 235.2777,
|
| 2911 |
+
"eval_samples_per_second": 873.479,
|
| 2912 |
+
"eval_steps_per_second": 54.595,
|
| 2913 |
+
"step": 2112000
|
| 2914 |
+
},
|
| 2915 |
+
{
|
| 2916 |
+
"epoch": 8.69,
|
| 2917 |
+
"eval_loss": 2.8771519660949707,
|
| 2918 |
+
"eval_runtime": 234.9372,
|
| 2919 |
+
"eval_samples_per_second": 874.744,
|
| 2920 |
+
"eval_steps_per_second": 54.674,
|
| 2921 |
+
"step": 2120000
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 8.72,
|
| 2925 |
+
"learning_rate": 1.1339948303176854e-06,
|
| 2926 |
+
"loss": 2.85,
|
| 2927 |
+
"step": 2128000
|
| 2928 |
+
},
|
| 2929 |
+
{
|
| 2930 |
+
"epoch": 8.72,
|
| 2931 |
+
"eval_loss": 2.8815581798553467,
|
| 2932 |
+
"eval_runtime": 233.3269,
|
| 2933 |
+
"eval_samples_per_second": 880.781,
|
| 2934 |
+
"eval_steps_per_second": 55.052,
|
| 2935 |
+
"step": 2128000
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 8.75,
|
| 2939 |
+
"eval_loss": 2.8723196983337402,
|
| 2940 |
+
"eval_runtime": 233.3963,
|
| 2941 |
+
"eval_samples_per_second": 880.52,
|
| 2942 |
+
"eval_steps_per_second": 55.035,
|
| 2943 |
+
"step": 2136000
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 8.79,
|
| 2947 |
+
"learning_rate": 1.067289252063704e-06,
|
| 2948 |
+
"loss": 2.8482,
|
| 2949 |
+
"step": 2144000
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 8.79,
|
| 2953 |
+
"eval_loss": 2.8833649158477783,
|
| 2954 |
+
"eval_runtime": 233.2426,
|
| 2955 |
+
"eval_samples_per_second": 881.1,
|
| 2956 |
+
"eval_steps_per_second": 55.071,
|
| 2957 |
+
"step": 2144000
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 8.82,
|
| 2961 |
+
"eval_loss": 2.8784103393554688,
|
| 2962 |
+
"eval_runtime": 232.8953,
|
| 2963 |
+
"eval_samples_per_second": 882.414,
|
| 2964 |
+
"eval_steps_per_second": 55.154,
|
| 2965 |
+
"step": 2152000
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 8.85,
|
| 2969 |
+
"learning_rate": 1.0005836738097224e-06,
|
| 2970 |
+
"loss": 2.8417,
|
| 2971 |
+
"step": 2160000
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 8.85,
|
| 2975 |
+
"eval_loss": 2.8758938312530518,
|
| 2976 |
+
"eval_runtime": 233.0635,
|
| 2977 |
+
"eval_samples_per_second": 881.777,
|
| 2978 |
+
"eval_steps_per_second": 55.114,
|
| 2979 |
+
"step": 2160000
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 8.88,
|
| 2983 |
+
"eval_loss": 2.881685256958008,
|
| 2984 |
+
"eval_runtime": 232.8737,
|
| 2985 |
+
"eval_samples_per_second": 882.495,
|
| 2986 |
+
"eval_steps_per_second": 55.159,
|
| 2987 |
+
"step": 2168000
|
| 2988 |
+
},
|
| 2989 |
+
{
|
| 2990 |
+
"epoch": 8.92,
|
| 2991 |
+
"learning_rate": 9.33878095555741e-07,
|
| 2992 |
+
"loss": 2.8336,
|
| 2993 |
+
"step": 2176000
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 8.92,
|
| 2997 |
+
"eval_loss": 2.8810818195343018,
|
| 2998 |
+
"eval_runtime": 233.1976,
|
| 2999 |
+
"eval_samples_per_second": 881.27,
|
| 3000 |
+
"eval_steps_per_second": 55.082,
|
| 3001 |
+
"step": 2176000
|
| 3002 |
+
},
|
| 3003 |
+
{
|
| 3004 |
+
"epoch": 8.95,
|
| 3005 |
+
"eval_loss": 2.8726937770843506,
|
| 3006 |
+
"eval_runtime": 232.1542,
|
| 3007 |
+
"eval_samples_per_second": 885.231,
|
| 3008 |
+
"eval_steps_per_second": 55.33,
|
| 3009 |
+
"step": 2184000
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 8.98,
|
| 3013 |
+
"learning_rate": 8.671725173017595e-07,
|
| 3014 |
+
"loss": 2.8514,
|
| 3015 |
+
"step": 2192000
|
| 3016 |
+
},
|
| 3017 |
+
{
|
| 3018 |
+
"epoch": 8.98,
|
| 3019 |
+
"eval_loss": 2.889397621154785,
|
| 3020 |
+
"eval_runtime": 232.9383,
|
| 3021 |
+
"eval_samples_per_second": 882.251,
|
| 3022 |
+
"eval_steps_per_second": 55.143,
|
| 3023 |
+
"step": 2192000
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 9.01,
|
| 3027 |
+
"eval_loss": 2.8750553131103516,
|
| 3028 |
+
"eval_runtime": 234.1088,
|
| 3029 |
+
"eval_samples_per_second": 877.84,
|
| 3030 |
+
"eval_steps_per_second": 54.868,
|
| 3031 |
+
"step": 2200000
|
| 3032 |
+
},
|
| 3033 |
+
{
|
| 3034 |
+
"epoch": 9.05,
|
| 3035 |
+
"learning_rate": 8.004669390477779e-07,
|
| 3036 |
+
"loss": 2.8312,
|
| 3037 |
+
"step": 2208000
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 9.05,
|
| 3041 |
+
"eval_loss": 2.877980947494507,
|
| 3042 |
+
"eval_runtime": 234.0757,
|
| 3043 |
+
"eval_samples_per_second": 877.964,
|
| 3044 |
+
"eval_steps_per_second": 54.875,
|
| 3045 |
+
"step": 2208000
|
| 3046 |
+
},
|
| 3047 |
+
{
|
| 3048 |
+
"epoch": 9.08,
|
| 3049 |
+
"eval_loss": 2.8862829208374023,
|
| 3050 |
+
"eval_runtime": 232.0198,
|
| 3051 |
+
"eval_samples_per_second": 885.743,
|
| 3052 |
+
"eval_steps_per_second": 55.362,
|
| 3053 |
+
"step": 2216000
|
| 3054 |
+
},
|
| 3055 |
+
{
|
| 3056 |
+
"epoch": 9.11,
|
| 3057 |
+
"learning_rate": 7.337613607937964e-07,
|
| 3058 |
+
"loss": 2.8315,
|
| 3059 |
+
"step": 2224000
|
| 3060 |
+
},
|
| 3061 |
+
{
|
| 3062 |
+
"epoch": 9.11,
|
| 3063 |
+
"eval_loss": 2.881166696548462,
|
| 3064 |
+
"eval_runtime": 232.6004,
|
| 3065 |
+
"eval_samples_per_second": 883.533,
|
| 3066 |
+
"eval_steps_per_second": 55.223,
|
| 3067 |
+
"step": 2224000
|
| 3068 |
+
},
|
| 3069 |
+
{
|
| 3070 |
+
"epoch": 9.15,
|
| 3071 |
+
"eval_loss": 2.871530532836914,
|
| 3072 |
+
"eval_runtime": 233.3854,
|
| 3073 |
+
"eval_samples_per_second": 880.561,
|
| 3074 |
+
"eval_steps_per_second": 55.038,
|
| 3075 |
+
"step": 2232000
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 9.18,
|
| 3079 |
+
"learning_rate": 6.67055782539815e-07,
|
| 3080 |
+
"loss": 2.8509,
|
| 3081 |
+
"step": 2240000
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 9.18,
|
| 3085 |
+
"eval_loss": 2.890825033187866,
|
| 3086 |
+
"eval_runtime": 233.6047,
|
| 3087 |
+
"eval_samples_per_second": 879.734,
|
| 3088 |
+
"eval_steps_per_second": 54.986,
|
| 3089 |
+
"step": 2240000
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 9.21,
|
| 3093 |
+
"eval_loss": 2.880777359008789,
|
| 3094 |
+
"eval_runtime": 232.8121,
|
| 3095 |
+
"eval_samples_per_second": 882.729,
|
| 3096 |
+
"eval_steps_per_second": 55.173,
|
| 3097 |
+
"step": 2248000
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 9.24,
|
| 3101 |
+
"learning_rate": 6.003502042858334e-07,
|
| 3102 |
+
"loss": 2.8394,
|
| 3103 |
+
"step": 2256000
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 9.24,
|
| 3107 |
+
"eval_loss": 2.8801651000976562,
|
| 3108 |
+
"eval_runtime": 235.058,
|
| 3109 |
+
"eval_samples_per_second": 874.295,
|
| 3110 |
+
"eval_steps_per_second": 54.646,
|
| 3111 |
+
"step": 2256000
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 9.28,
|
| 3115 |
+
"eval_loss": 2.8692407608032227,
|
| 3116 |
+
"eval_runtime": 234.8843,
|
| 3117 |
+
"eval_samples_per_second": 874.941,
|
| 3118 |
+
"eval_steps_per_second": 54.686,
|
| 3119 |
+
"step": 2264000
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 9.31,
|
| 3123 |
+
"learning_rate": 5.33644626031852e-07,
|
| 3124 |
+
"loss": 2.8332,
|
| 3125 |
+
"step": 2272000
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 9.31,
|
| 3129 |
+
"eval_loss": 2.871225357055664,
|
| 3130 |
+
"eval_runtime": 233.9349,
|
| 3131 |
+
"eval_samples_per_second": 878.492,
|
| 3132 |
+
"eval_steps_per_second": 54.908,
|
| 3133 |
+
"step": 2272000
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 9.34,
|
| 3137 |
+
"eval_loss": 2.8688416481018066,
|
| 3138 |
+
"eval_runtime": 233.3243,
|
| 3139 |
+
"eval_samples_per_second": 880.791,
|
| 3140 |
+
"eval_steps_per_second": 55.052,
|
| 3141 |
+
"step": 2280000
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 9.38,
|
| 3145 |
+
"learning_rate": 4.669390477778705e-07,
|
| 3146 |
+
"loss": 2.837,
|
| 3147 |
+
"step": 2288000
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 9.38,
|
| 3151 |
+
"eval_loss": 2.8778717517852783,
|
| 3152 |
+
"eval_runtime": 233.117,
|
| 3153 |
+
"eval_samples_per_second": 881.574,
|
| 3154 |
+
"eval_steps_per_second": 55.101,
|
| 3155 |
+
"step": 2288000
|
| 3156 |
+
},
|
| 3157 |
+
{
|
| 3158 |
+
"epoch": 9.41,
|
| 3159 |
+
"eval_loss": 2.879408597946167,
|
| 3160 |
+
"eval_runtime": 233.2294,
|
| 3161 |
+
"eval_samples_per_second": 881.149,
|
| 3162 |
+
"eval_steps_per_second": 55.075,
|
| 3163 |
+
"step": 2296000
|
| 3164 |
+
},
|
| 3165 |
+
{
|
| 3166 |
+
"epoch": 9.44,
|
| 3167 |
+
"learning_rate": 4.0023346952388894e-07,
|
| 3168 |
+
"loss": 2.8344,
|
| 3169 |
+
"step": 2304000
|
| 3170 |
+
},
|
| 3171 |
+
{
|
| 3172 |
+
"epoch": 9.44,
|
| 3173 |
+
"eval_loss": 2.8751115798950195,
|
| 3174 |
+
"eval_runtime": 233.3516,
|
| 3175 |
+
"eval_samples_per_second": 880.688,
|
| 3176 |
+
"eval_steps_per_second": 55.046,
|
| 3177 |
+
"step": 2304000
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 9.47,
|
| 3181 |
+
"eval_loss": 2.874969482421875,
|
| 3182 |
+
"eval_runtime": 233.1427,
|
| 3183 |
+
"eval_samples_per_second": 881.477,
|
| 3184 |
+
"eval_steps_per_second": 55.095,
|
| 3185 |
+
"step": 2312000
|
| 3186 |
+
},
|
| 3187 |
+
{
|
| 3188 |
+
"epoch": 9.51,
|
| 3189 |
+
"learning_rate": 3.335278912699075e-07,
|
| 3190 |
+
"loss": 2.8376,
|
| 3191 |
+
"step": 2320000
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 9.51,
|
| 3195 |
+
"eval_loss": 2.8837876319885254,
|
| 3196 |
+
"eval_runtime": 233.3471,
|
| 3197 |
+
"eval_samples_per_second": 880.705,
|
| 3198 |
+
"eval_steps_per_second": 55.047,
|
| 3199 |
+
"step": 2320000
|
| 3200 |
+
},
|
| 3201 |
+
{
|
| 3202 |
+
"epoch": 9.54,
|
| 3203 |
+
"eval_loss": 2.8825130462646484,
|
| 3204 |
+
"eval_runtime": 233.8169,
|
| 3205 |
+
"eval_samples_per_second": 878.935,
|
| 3206 |
+
"eval_steps_per_second": 54.936,
|
| 3207 |
+
"step": 2328000
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 9.57,
|
| 3211 |
+
"learning_rate": 2.66822313015926e-07,
|
| 3212 |
+
"loss": 2.8329,
|
| 3213 |
+
"step": 2336000
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 9.57,
|
| 3217 |
+
"eval_loss": 2.880941390991211,
|
| 3218 |
+
"eval_runtime": 234.5491,
|
| 3219 |
+
"eval_samples_per_second": 876.192,
|
| 3220 |
+
"eval_steps_per_second": 54.765,
|
| 3221 |
+
"step": 2336000
|
| 3222 |
+
},
|
| 3223 |
+
{
|
| 3224 |
+
"epoch": 9.6,
|
| 3225 |
+
"eval_loss": 2.8842968940734863,
|
| 3226 |
+
"eval_runtime": 233.344,
|
| 3227 |
+
"eval_samples_per_second": 880.717,
|
| 3228 |
+
"eval_steps_per_second": 55.047,
|
| 3229 |
+
"step": 2344000
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 9.64,
|
| 3233 |
+
"learning_rate": 2.0011673476194447e-07,
|
| 3234 |
+
"loss": 2.8264,
|
| 3235 |
+
"step": 2352000
|
| 3236 |
+
},
|
| 3237 |
+
{
|
| 3238 |
+
"epoch": 9.64,
|
| 3239 |
+
"eval_loss": 2.8784091472625732,
|
| 3240 |
+
"eval_runtime": 233.7157,
|
| 3241 |
+
"eval_samples_per_second": 879.316,
|
| 3242 |
+
"eval_steps_per_second": 54.96,
|
| 3243 |
+
"step": 2352000
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 9.67,
|
| 3247 |
+
"eval_loss": 2.8687503337860107,
|
| 3248 |
+
"eval_runtime": 233.7026,
|
| 3249 |
+
"eval_samples_per_second": 879.365,
|
| 3250 |
+
"eval_steps_per_second": 54.963,
|
| 3251 |
+
"step": 2360000
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 9.7,
|
| 3255 |
+
"learning_rate": 1.33411156507963e-07,
|
| 3256 |
+
"loss": 2.8317,
|
| 3257 |
+
"step": 2368000
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 9.7,
|
| 3261 |
+
"eval_loss": 2.8792617321014404,
|
| 3262 |
+
"eval_runtime": 235.256,
|
| 3263 |
+
"eval_samples_per_second": 873.559,
|
| 3264 |
+
"eval_steps_per_second": 54.6,
|
| 3265 |
+
"step": 2368000
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 9.74,
|
| 3269 |
+
"eval_loss": 2.8814611434936523,
|
| 3270 |
+
"eval_runtime": 235.5823,
|
| 3271 |
+
"eval_samples_per_second": 872.349,
|
| 3272 |
+
"eval_steps_per_second": 54.524,
|
| 3273 |
+
"step": 2376000
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 9.77,
|
| 3277 |
+
"learning_rate": 6.67055782539815e-08,
|
| 3278 |
+
"loss": 2.8328,
|
| 3279 |
+
"step": 2384000
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 9.77,
|
| 3283 |
+
"eval_loss": 2.8755900859832764,
|
| 3284 |
+
"eval_runtime": 233.6052,
|
| 3285 |
+
"eval_samples_per_second": 879.732,
|
| 3286 |
+
"eval_steps_per_second": 54.986,
|
| 3287 |
+
"step": 2384000
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 9.8,
|
| 3291 |
+
"eval_loss": 2.869050979614258,
|
| 3292 |
+
"eval_runtime": 233.5136,
|
| 3293 |
+
"eval_samples_per_second": 880.077,
|
| 3294 |
+
"eval_steps_per_second": 55.007,
|
| 3295 |
+
"step": 2392000
|
| 3296 |
+
},
|
| 3297 |
+
{
|
| 3298 |
+
"epoch": 9.83,
|
| 3299 |
+
"learning_rate": 0.0,
|
| 3300 |
+
"loss": 2.841,
|
| 3301 |
+
"step": 2400000
|
| 3302 |
+
},
|
| 3303 |
+
{
|
| 3304 |
+
"epoch": 9.83,
|
| 3305 |
+
"eval_loss": 2.867439031600952,
|
| 3306 |
+
"eval_runtime": 233.911,
|
| 3307 |
+
"eval_samples_per_second": 878.582,
|
| 3308 |
+
"eval_steps_per_second": 54.914,
|
| 3309 |
+
"step": 2400000
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 9.83,
|
| 3313 |
+
"step": 2400000,
|
| 3314 |
+
"total_flos": 7.305293129309786e+17,
|
| 3315 |
+
"train_loss": 2.773840421549479,
|
| 3316 |
+
"train_runtime": 219744.0708,
|
| 3317 |
+
"train_samples_per_second": 174.749,
|
| 3318 |
+
"train_steps_per_second": 10.922
|
| 3319 |
+
}
|
| 3320 |
+
],
|
| 3321 |
+
"logging_steps": 16000,
|
| 3322 |
+
"max_steps": 2400000,
|
| 3323 |
+
"num_train_epochs": 10,
|
| 3324 |
+
"save_steps": 32000,
|
| 3325 |
+
"total_flos": 7.305293129309786e+17,
|
| 3326 |
+
"trial_name": null,
|
| 3327 |
+
"trial_params": null
|
| 3328 |
+
}
|