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 [DouglasPontes/2020-Q3-50p-filtered-random](https://huggingface.co/DouglasPontes/2020-Q3-50p-filtered-random) 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 [DouglasPontes/2020-Q3-50p-filtered-random](https://huggingface.co/DouglasPontes/2020-Q3-50p-filtered-random) on an unknown dataset.
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
+
- Loss: 2.2730
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.11,
|
| 3 |
+
"eval_loss": 2.2730469703674316,
|
| 4 |
+
"eval_runtime": 247.9141,
|
| 5 |
+
"eval_samples": 200000,
|
| 6 |
+
"eval_samples_per_second": 806.731,
|
| 7 |
+
"eval_steps_per_second": 50.421,
|
| 8 |
+
"perplexity": 9.708938639114127,
|
| 9 |
+
"train_loss": 2.4345372688802085,
|
| 10 |
+
"train_runtime": 232385.3236,
|
| 11 |
+
"train_samples": 3800001,
|
| 12 |
+
"train_samples_per_second": 165.243,
|
| 13 |
+
"train_steps_per_second": 10.328
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.11,
|
| 3 |
+
"eval_loss": 2.2730469703674316,
|
| 4 |
+
"eval_runtime": 247.9141,
|
| 5 |
+
"eval_samples": 200000,
|
| 6 |
+
"eval_samples_per_second": 806.731,
|
| 7 |
+
"eval_steps_per_second": 50.421,
|
| 8 |
+
"perplexity": 9.708938639114127
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.11,
|
| 3 |
+
"train_loss": 2.4345372688802085,
|
| 4 |
+
"train_runtime": 232385.3236,
|
| 5 |
+
"train_samples": 3800001,
|
| 6 |
+
"train_samples_per_second": 165.243,
|
| 7 |
+
"train_steps_per_second": 10.328
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3328 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 2.270662546157837,
|
| 3 |
+
"best_model_checkpoint": "./model_tweets_2020_Q4_50_rand/checkpoint-2240000",
|
| 4 |
+
"epoch": 10.105220609597433,
|
| 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": 2.3816494941711426,
|
| 14 |
+
"eval_runtime": 236.9637,
|
| 15 |
+
"eval_samples_per_second": 844.011,
|
| 16 |
+
"eval_steps_per_second": 52.751,
|
| 17 |
+
"step": 8000
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.07,
|
| 21 |
+
"learning_rate": 4.0726666666666665e-07,
|
| 22 |
+
"loss": 2.5089,
|
| 23 |
+
"step": 16000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.07,
|
| 27 |
+
"eval_loss": 2.364738702774048,
|
| 28 |
+
"eval_runtime": 237.3943,
|
| 29 |
+
"eval_samples_per_second": 842.48,
|
| 30 |
+
"eval_steps_per_second": 52.655,
|
| 31 |
+
"step": 16000
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.1,
|
| 35 |
+
"eval_loss": 2.355194091796875,
|
| 36 |
+
"eval_runtime": 237.9121,
|
| 37 |
+
"eval_samples_per_second": 840.647,
|
| 38 |
+
"eval_steps_per_second": 52.54,
|
| 39 |
+
"step": 24000
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.13,
|
| 43 |
+
"learning_rate": 4.0453333333333336e-07,
|
| 44 |
+
"loss": 2.4989,
|
| 45 |
+
"step": 32000
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.13,
|
| 49 |
+
"eval_loss": 2.3528289794921875,
|
| 50 |
+
"eval_runtime": 238.6586,
|
| 51 |
+
"eval_samples_per_second": 838.017,
|
| 52 |
+
"eval_steps_per_second": 52.376,
|
| 53 |
+
"step": 32000
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.17,
|
| 57 |
+
"eval_loss": 2.3485841751098633,
|
| 58 |
+
"eval_runtime": 239.5908,
|
| 59 |
+
"eval_samples_per_second": 834.756,
|
| 60 |
+
"eval_steps_per_second": 52.172,
|
| 61 |
+
"step": 40000
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.2,
|
| 65 |
+
"learning_rate": 4.018e-07,
|
| 66 |
+
"loss": 2.4836,
|
| 67 |
+
"step": 48000
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.2,
|
| 71 |
+
"eval_loss": 2.34626841545105,
|
| 72 |
+
"eval_runtime": 238.7656,
|
| 73 |
+
"eval_samples_per_second": 837.642,
|
| 74 |
+
"eval_steps_per_second": 52.353,
|
| 75 |
+
"step": 48000
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.24,
|
| 79 |
+
"eval_loss": 2.3410613536834717,
|
| 80 |
+
"eval_runtime": 240.0423,
|
| 81 |
+
"eval_samples_per_second": 833.186,
|
| 82 |
+
"eval_steps_per_second": 52.074,
|
| 83 |
+
"step": 56000
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.27,
|
| 87 |
+
"learning_rate": 3.9906666666666667e-07,
|
| 88 |
+
"loss": 2.4904,
|
| 89 |
+
"step": 64000
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.27,
|
| 93 |
+
"eval_loss": 2.3394079208374023,
|
| 94 |
+
"eval_runtime": 239.0867,
|
| 95 |
+
"eval_samples_per_second": 836.517,
|
| 96 |
+
"eval_steps_per_second": 52.282,
|
| 97 |
+
"step": 64000
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.3,
|
| 101 |
+
"eval_loss": 2.3349857330322266,
|
| 102 |
+
"eval_runtime": 238.1477,
|
| 103 |
+
"eval_samples_per_second": 839.815,
|
| 104 |
+
"eval_steps_per_second": 52.488,
|
| 105 |
+
"step": 72000
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.34,
|
| 109 |
+
"learning_rate": 3.963333333333333e-07,
|
| 110 |
+
"loss": 2.4733,
|
| 111 |
+
"step": 80000
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.34,
|
| 115 |
+
"eval_loss": 2.330885410308838,
|
| 116 |
+
"eval_runtime": 239.5566,
|
| 117 |
+
"eval_samples_per_second": 834.876,
|
| 118 |
+
"eval_steps_per_second": 52.18,
|
| 119 |
+
"step": 80000
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.37,
|
| 123 |
+
"eval_loss": 2.3288557529449463,
|
| 124 |
+
"eval_runtime": 239.3525,
|
| 125 |
+
"eval_samples_per_second": 835.588,
|
| 126 |
+
"eval_steps_per_second": 52.224,
|
| 127 |
+
"step": 88000
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 0.4,
|
| 131 |
+
"learning_rate": 3.936e-07,
|
| 132 |
+
"loss": 2.4675,
|
| 133 |
+
"step": 96000
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 0.4,
|
| 137 |
+
"eval_loss": 2.3381102085113525,
|
| 138 |
+
"eval_runtime": 239.1998,
|
| 139 |
+
"eval_samples_per_second": 836.121,
|
| 140 |
+
"eval_steps_per_second": 52.258,
|
| 141 |
+
"step": 96000
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.44,
|
| 145 |
+
"eval_loss": 2.331658363342285,
|
| 146 |
+
"eval_runtime": 237.7084,
|
| 147 |
+
"eval_samples_per_second": 841.367,
|
| 148 |
+
"eval_steps_per_second": 52.585,
|
| 149 |
+
"step": 104000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.47,
|
| 153 |
+
"learning_rate": 3.908666666666667e-07,
|
| 154 |
+
"loss": 2.4762,
|
| 155 |
+
"step": 112000
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.47,
|
| 159 |
+
"eval_loss": 2.3218166828155518,
|
| 160 |
+
"eval_runtime": 239.0796,
|
| 161 |
+
"eval_samples_per_second": 836.541,
|
| 162 |
+
"eval_steps_per_second": 52.284,
|
| 163 |
+
"step": 112000
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.51,
|
| 167 |
+
"eval_loss": 2.326749086380005,
|
| 168 |
+
"eval_runtime": 239.0317,
|
| 169 |
+
"eval_samples_per_second": 836.709,
|
| 170 |
+
"eval_steps_per_second": 52.294,
|
| 171 |
+
"step": 120000
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.54,
|
| 175 |
+
"learning_rate": 3.8813333333333334e-07,
|
| 176 |
+
"loss": 2.4616,
|
| 177 |
+
"step": 128000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.54,
|
| 181 |
+
"eval_loss": 2.3240551948547363,
|
| 182 |
+
"eval_runtime": 238.0739,
|
| 183 |
+
"eval_samples_per_second": 840.075,
|
| 184 |
+
"eval_steps_per_second": 52.505,
|
| 185 |
+
"step": 128000
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.57,
|
| 189 |
+
"eval_loss": 2.3280670642852783,
|
| 190 |
+
"eval_runtime": 238.6608,
|
| 191 |
+
"eval_samples_per_second": 838.009,
|
| 192 |
+
"eval_steps_per_second": 52.376,
|
| 193 |
+
"step": 136000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.61,
|
| 197 |
+
"learning_rate": 3.854e-07,
|
| 198 |
+
"loss": 2.4601,
|
| 199 |
+
"step": 144000
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.61,
|
| 203 |
+
"eval_loss": 2.315213680267334,
|
| 204 |
+
"eval_runtime": 238.6484,
|
| 205 |
+
"eval_samples_per_second": 838.053,
|
| 206 |
+
"eval_steps_per_second": 52.378,
|
| 207 |
+
"step": 144000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.64,
|
| 211 |
+
"eval_loss": 2.3242688179016113,
|
| 212 |
+
"eval_runtime": 239.0187,
|
| 213 |
+
"eval_samples_per_second": 836.755,
|
| 214 |
+
"eval_steps_per_second": 52.297,
|
| 215 |
+
"step": 152000
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.67,
|
| 219 |
+
"learning_rate": 3.8266666666666665e-07,
|
| 220 |
+
"loss": 2.4563,
|
| 221 |
+
"step": 160000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.67,
|
| 225 |
+
"eval_loss": 2.3201961517333984,
|
| 226 |
+
"eval_runtime": 240.1294,
|
| 227 |
+
"eval_samples_per_second": 832.884,
|
| 228 |
+
"eval_steps_per_second": 52.055,
|
| 229 |
+
"step": 160000
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.71,
|
| 233 |
+
"eval_loss": 2.3193206787109375,
|
| 234 |
+
"eval_runtime": 241.1825,
|
| 235 |
+
"eval_samples_per_second": 829.247,
|
| 236 |
+
"eval_steps_per_second": 51.828,
|
| 237 |
+
"step": 168000
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.74,
|
| 241 |
+
"learning_rate": 3.799333333333333e-07,
|
| 242 |
+
"loss": 2.459,
|
| 243 |
+
"step": 176000
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.74,
|
| 247 |
+
"eval_loss": 2.318166732788086,
|
| 248 |
+
"eval_runtime": 238.5889,
|
| 249 |
+
"eval_samples_per_second": 838.262,
|
| 250 |
+
"eval_steps_per_second": 52.391,
|
| 251 |
+
"step": 176000
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.77,
|
| 255 |
+
"eval_loss": 2.324726104736328,
|
| 256 |
+
"eval_runtime": 238.4567,
|
| 257 |
+
"eval_samples_per_second": 838.727,
|
| 258 |
+
"eval_steps_per_second": 52.42,
|
| 259 |
+
"step": 184000
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.81,
|
| 263 |
+
"learning_rate": 3.772e-07,
|
| 264 |
+
"loss": 2.4639,
|
| 265 |
+
"step": 192000
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.81,
|
| 269 |
+
"eval_loss": 2.320105791091919,
|
| 270 |
+
"eval_runtime": 238.6268,
|
| 271 |
+
"eval_samples_per_second": 838.129,
|
| 272 |
+
"eval_steps_per_second": 52.383,
|
| 273 |
+
"step": 192000
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.84,
|
| 277 |
+
"eval_loss": 2.3243095874786377,
|
| 278 |
+
"eval_runtime": 239.9922,
|
| 279 |
+
"eval_samples_per_second": 833.36,
|
| 280 |
+
"eval_steps_per_second": 52.085,
|
| 281 |
+
"step": 200000
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.88,
|
| 285 |
+
"learning_rate": 3.7446666666666667e-07,
|
| 286 |
+
"loss": 2.4561,
|
| 287 |
+
"step": 208000
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.88,
|
| 291 |
+
"eval_loss": 2.321760416030884,
|
| 292 |
+
"eval_runtime": 239.3254,
|
| 293 |
+
"eval_samples_per_second": 835.682,
|
| 294 |
+
"eval_steps_per_second": 52.23,
|
| 295 |
+
"step": 208000
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.91,
|
| 299 |
+
"eval_loss": 2.3137271404266357,
|
| 300 |
+
"eval_runtime": 240.8808,
|
| 301 |
+
"eval_samples_per_second": 830.286,
|
| 302 |
+
"eval_steps_per_second": 51.893,
|
| 303 |
+
"step": 216000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.94,
|
| 307 |
+
"learning_rate": 3.7173333333333333e-07,
|
| 308 |
+
"loss": 2.4556,
|
| 309 |
+
"step": 224000
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.94,
|
| 313 |
+
"eval_loss": 2.318033456802368,
|
| 314 |
+
"eval_runtime": 239.3207,
|
| 315 |
+
"eval_samples_per_second": 835.699,
|
| 316 |
+
"eval_steps_per_second": 52.231,
|
| 317 |
+
"step": 224000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.98,
|
| 321 |
+
"eval_loss": 2.314727783203125,
|
| 322 |
+
"eval_runtime": 238.4213,
|
| 323 |
+
"eval_samples_per_second": 838.851,
|
| 324 |
+
"eval_steps_per_second": 52.428,
|
| 325 |
+
"step": 232000
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 1.01,
|
| 329 |
+
"learning_rate": 3.69e-07,
|
| 330 |
+
"loss": 2.4573,
|
| 331 |
+
"step": 240000
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 1.01,
|
| 335 |
+
"eval_loss": 2.309976100921631,
|
| 336 |
+
"eval_runtime": 238.672,
|
| 337 |
+
"eval_samples_per_second": 837.97,
|
| 338 |
+
"eval_steps_per_second": 52.373,
|
| 339 |
+
"step": 240000
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.04,
|
| 343 |
+
"eval_loss": 2.311811923980713,
|
| 344 |
+
"eval_runtime": 239.3065,
|
| 345 |
+
"eval_samples_per_second": 835.748,
|
| 346 |
+
"eval_steps_per_second": 52.234,
|
| 347 |
+
"step": 248000
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 1.08,
|
| 351 |
+
"learning_rate": 3.6626666666666664e-07,
|
| 352 |
+
"loss": 2.4516,
|
| 353 |
+
"step": 256000
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.08,
|
| 357 |
+
"eval_loss": 2.315763235092163,
|
| 358 |
+
"eval_runtime": 239.6224,
|
| 359 |
+
"eval_samples_per_second": 834.646,
|
| 360 |
+
"eval_steps_per_second": 52.165,
|
| 361 |
+
"step": 256000
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 1.11,
|
| 365 |
+
"eval_loss": 2.3132565021514893,
|
| 366 |
+
"eval_runtime": 241.721,
|
| 367 |
+
"eval_samples_per_second": 827.4,
|
| 368 |
+
"eval_steps_per_second": 51.713,
|
| 369 |
+
"step": 264000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.15,
|
| 373 |
+
"learning_rate": 3.6353333333333335e-07,
|
| 374 |
+
"loss": 2.4561,
|
| 375 |
+
"step": 272000
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 1.15,
|
| 379 |
+
"eval_loss": 2.3065385818481445,
|
| 380 |
+
"eval_runtime": 241.1529,
|
| 381 |
+
"eval_samples_per_second": 829.349,
|
| 382 |
+
"eval_steps_per_second": 51.834,
|
| 383 |
+
"step": 272000
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.18,
|
| 387 |
+
"eval_loss": 2.318328857421875,
|
| 388 |
+
"eval_runtime": 240.9143,
|
| 389 |
+
"eval_samples_per_second": 830.171,
|
| 390 |
+
"eval_steps_per_second": 51.886,
|
| 391 |
+
"step": 280000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 1.21,
|
| 395 |
+
"learning_rate": 3.608e-07,
|
| 396 |
+
"loss": 2.4476,
|
| 397 |
+
"step": 288000
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 1.21,
|
| 401 |
+
"eval_loss": 2.310638666152954,
|
| 402 |
+
"eval_runtime": 240.8865,
|
| 403 |
+
"eval_samples_per_second": 830.267,
|
| 404 |
+
"eval_steps_per_second": 51.892,
|
| 405 |
+
"step": 288000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.25,
|
| 409 |
+
"eval_loss": 2.3131144046783447,
|
| 410 |
+
"eval_runtime": 238.8251,
|
| 411 |
+
"eval_samples_per_second": 837.433,
|
| 412 |
+
"eval_steps_per_second": 52.34,
|
| 413 |
+
"step": 296000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.28,
|
| 417 |
+
"learning_rate": 3.5806666666666666e-07,
|
| 418 |
+
"loss": 2.4503,
|
| 419 |
+
"step": 304000
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.28,
|
| 423 |
+
"eval_loss": 2.3103673458099365,
|
| 424 |
+
"eval_runtime": 238.9913,
|
| 425 |
+
"eval_samples_per_second": 836.851,
|
| 426 |
+
"eval_steps_per_second": 52.303,
|
| 427 |
+
"step": 304000
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 1.31,
|
| 431 |
+
"eval_loss": 2.310051679611206,
|
| 432 |
+
"eval_runtime": 239.4298,
|
| 433 |
+
"eval_samples_per_second": 835.318,
|
| 434 |
+
"eval_steps_per_second": 52.207,
|
| 435 |
+
"step": 312000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 1.35,
|
| 439 |
+
"learning_rate": 3.553333333333333e-07,
|
| 440 |
+
"loss": 2.4495,
|
| 441 |
+
"step": 320000
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.35,
|
| 445 |
+
"eval_loss": 2.3085968494415283,
|
| 446 |
+
"eval_runtime": 239.7653,
|
| 447 |
+
"eval_samples_per_second": 834.149,
|
| 448 |
+
"eval_steps_per_second": 52.134,
|
| 449 |
+
"step": 320000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.38,
|
| 453 |
+
"eval_loss": 2.3057291507720947,
|
| 454 |
+
"eval_runtime": 240.7073,
|
| 455 |
+
"eval_samples_per_second": 830.885,
|
| 456 |
+
"eval_steps_per_second": 51.93,
|
| 457 |
+
"step": 328000
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 1.41,
|
| 461 |
+
"learning_rate": 3.5259999999999997e-07,
|
| 462 |
+
"loss": 2.4534,
|
| 463 |
+
"step": 336000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.41,
|
| 467 |
+
"eval_loss": 2.3085904121398926,
|
| 468 |
+
"eval_runtime": 240.303,
|
| 469 |
+
"eval_samples_per_second": 832.282,
|
| 470 |
+
"eval_steps_per_second": 52.018,
|
| 471 |
+
"step": 336000
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 1.45,
|
| 475 |
+
"eval_loss": 2.309332847595215,
|
| 476 |
+
"eval_runtime": 240.8676,
|
| 477 |
+
"eval_samples_per_second": 830.332,
|
| 478 |
+
"eval_steps_per_second": 51.896,
|
| 479 |
+
"step": 344000
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.48,
|
| 483 |
+
"learning_rate": 3.498666666666667e-07,
|
| 484 |
+
"loss": 2.4486,
|
| 485 |
+
"step": 352000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.48,
|
| 489 |
+
"eval_loss": 2.3018343448638916,
|
| 490 |
+
"eval_runtime": 241.3997,
|
| 491 |
+
"eval_samples_per_second": 828.501,
|
| 492 |
+
"eval_steps_per_second": 51.781,
|
| 493 |
+
"step": 352000
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.52,
|
| 497 |
+
"eval_loss": 2.305995464324951,
|
| 498 |
+
"eval_runtime": 239.2635,
|
| 499 |
+
"eval_samples_per_second": 835.898,
|
| 500 |
+
"eval_steps_per_second": 52.244,
|
| 501 |
+
"step": 360000
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.55,
|
| 505 |
+
"learning_rate": 3.4713333333333333e-07,
|
| 506 |
+
"loss": 2.4457,
|
| 507 |
+
"step": 368000
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.55,
|
| 511 |
+
"eval_loss": 2.3082966804504395,
|
| 512 |
+
"eval_runtime": 239.5152,
|
| 513 |
+
"eval_samples_per_second": 835.02,
|
| 514 |
+
"eval_steps_per_second": 52.189,
|
| 515 |
+
"step": 368000
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.58,
|
| 519 |
+
"eval_loss": 2.3109591007232666,
|
| 520 |
+
"eval_runtime": 240.8072,
|
| 521 |
+
"eval_samples_per_second": 830.54,
|
| 522 |
+
"eval_steps_per_second": 51.909,
|
| 523 |
+
"step": 376000
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 1.62,
|
| 527 |
+
"learning_rate": 3.444e-07,
|
| 528 |
+
"loss": 2.4443,
|
| 529 |
+
"step": 384000
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.62,
|
| 533 |
+
"eval_loss": 2.297455310821533,
|
| 534 |
+
"eval_runtime": 240.6668,
|
| 535 |
+
"eval_samples_per_second": 831.024,
|
| 536 |
+
"eval_steps_per_second": 51.939,
|
| 537 |
+
"step": 384000
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 1.65,
|
| 541 |
+
"eval_loss": 2.300872564315796,
|
| 542 |
+
"eval_runtime": 240.6097,
|
| 543 |
+
"eval_samples_per_second": 831.222,
|
| 544 |
+
"eval_steps_per_second": 51.951,
|
| 545 |
+
"step": 392000
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.68,
|
| 549 |
+
"learning_rate": 3.416666666666667e-07,
|
| 550 |
+
"loss": 2.4405,
|
| 551 |
+
"step": 400000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 1.68,
|
| 555 |
+
"eval_loss": 2.3067097663879395,
|
| 556 |
+
"eval_runtime": 240.2686,
|
| 557 |
+
"eval_samples_per_second": 832.402,
|
| 558 |
+
"eval_steps_per_second": 52.025,
|
| 559 |
+
"step": 400000
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 1.72,
|
| 563 |
+
"eval_loss": 2.30265212059021,
|
| 564 |
+
"eval_runtime": 240.64,
|
| 565 |
+
"eval_samples_per_second": 831.117,
|
| 566 |
+
"eval_steps_per_second": 51.945,
|
| 567 |
+
"step": 408000
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.75,
|
| 571 |
+
"learning_rate": 3.3893333333333335e-07,
|
| 572 |
+
"loss": 2.4531,
|
| 573 |
+
"step": 416000
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 1.75,
|
| 577 |
+
"eval_loss": 2.3050363063812256,
|
| 578 |
+
"eval_runtime": 240.3135,
|
| 579 |
+
"eval_samples_per_second": 832.246,
|
| 580 |
+
"eval_steps_per_second": 52.015,
|
| 581 |
+
"step": 416000
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 1.79,
|
| 585 |
+
"eval_loss": 2.302565574645996,
|
| 586 |
+
"eval_runtime": 239.3844,
|
| 587 |
+
"eval_samples_per_second": 835.476,
|
| 588 |
+
"eval_steps_per_second": 52.217,
|
| 589 |
+
"step": 424000
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.82,
|
| 593 |
+
"learning_rate": 3.3619999999999995e-07,
|
| 594 |
+
"loss": 2.4539,
|
| 595 |
+
"step": 432000
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 1.82,
|
| 599 |
+
"eval_loss": 2.2928755283355713,
|
| 600 |
+
"eval_runtime": 241.5914,
|
| 601 |
+
"eval_samples_per_second": 827.844,
|
| 602 |
+
"eval_steps_per_second": 51.74,
|
| 603 |
+
"step": 432000
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.85,
|
| 607 |
+
"eval_loss": 2.305102825164795,
|
| 608 |
+
"eval_runtime": 241.5249,
|
| 609 |
+
"eval_samples_per_second": 828.072,
|
| 610 |
+
"eval_steps_per_second": 51.754,
|
| 611 |
+
"step": 440000
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.89,
|
| 615 |
+
"learning_rate": 3.3346666666666666e-07,
|
| 616 |
+
"loss": 2.4499,
|
| 617 |
+
"step": 448000
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 1.89,
|
| 621 |
+
"eval_loss": 2.3034677505493164,
|
| 622 |
+
"eval_runtime": 241.4767,
|
| 623 |
+
"eval_samples_per_second": 828.237,
|
| 624 |
+
"eval_steps_per_second": 51.765,
|
| 625 |
+
"step": 448000
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 1.92,
|
| 629 |
+
"eval_loss": 2.301123857498169,
|
| 630 |
+
"eval_runtime": 239.2116,
|
| 631 |
+
"eval_samples_per_second": 836.08,
|
| 632 |
+
"eval_steps_per_second": 52.255,
|
| 633 |
+
"step": 456000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.95,
|
| 637 |
+
"learning_rate": 3.307333333333333e-07,
|
| 638 |
+
"loss": 2.4401,
|
| 639 |
+
"step": 464000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.95,
|
| 643 |
+
"eval_loss": 2.291990280151367,
|
| 644 |
+
"eval_runtime": 241.865,
|
| 645 |
+
"eval_samples_per_second": 826.907,
|
| 646 |
+
"eval_steps_per_second": 51.682,
|
| 647 |
+
"step": 464000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.99,
|
| 651 |
+
"eval_loss": 2.299881935119629,
|
| 652 |
+
"eval_runtime": 239.8579,
|
| 653 |
+
"eval_samples_per_second": 833.827,
|
| 654 |
+
"eval_steps_per_second": 52.114,
|
| 655 |
+
"step": 472000
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 2.02,
|
| 659 |
+
"learning_rate": 3.28e-07,
|
| 660 |
+
"loss": 2.4401,
|
| 661 |
+
"step": 480000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 2.02,
|
| 665 |
+
"eval_loss": 2.303424119949341,
|
| 666 |
+
"eval_runtime": 240.1835,
|
| 667 |
+
"eval_samples_per_second": 832.697,
|
| 668 |
+
"eval_steps_per_second": 52.044,
|
| 669 |
+
"step": 480000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.05,
|
| 673 |
+
"eval_loss": 2.30208683013916,
|
| 674 |
+
"eval_runtime": 239.7131,
|
| 675 |
+
"eval_samples_per_second": 834.331,
|
| 676 |
+
"eval_steps_per_second": 52.146,
|
| 677 |
+
"step": 488000
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 2.09,
|
| 681 |
+
"learning_rate": 3.252666666666667e-07,
|
| 682 |
+
"loss": 2.4433,
|
| 683 |
+
"step": 496000
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 2.09,
|
| 687 |
+
"eval_loss": 2.3102383613586426,
|
| 688 |
+
"eval_runtime": 239.6539,
|
| 689 |
+
"eval_samples_per_second": 834.537,
|
| 690 |
+
"eval_steps_per_second": 52.159,
|
| 691 |
+
"step": 496000
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 2.12,
|
| 695 |
+
"eval_loss": 2.2985267639160156,
|
| 696 |
+
"eval_runtime": 239.3303,
|
| 697 |
+
"eval_samples_per_second": 835.665,
|
| 698 |
+
"eval_steps_per_second": 52.229,
|
| 699 |
+
"step": 504000
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 2.16,
|
| 703 |
+
"learning_rate": 3.2253333333333334e-07,
|
| 704 |
+
"loss": 2.4445,
|
| 705 |
+
"step": 512000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 2.16,
|
| 709 |
+
"eval_loss": 2.3018200397491455,
|
| 710 |
+
"eval_runtime": 239.5684,
|
| 711 |
+
"eval_samples_per_second": 834.835,
|
| 712 |
+
"eval_steps_per_second": 52.177,
|
| 713 |
+
"step": 512000
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 2.19,
|
| 717 |
+
"eval_loss": 2.2995855808258057,
|
| 718 |
+
"eval_runtime": 242.6094,
|
| 719 |
+
"eval_samples_per_second": 824.37,
|
| 720 |
+
"eval_steps_per_second": 51.523,
|
| 721 |
+
"step": 520000
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 2.22,
|
| 725 |
+
"learning_rate": 3.198e-07,
|
| 726 |
+
"loss": 2.4379,
|
| 727 |
+
"step": 528000
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 2.22,
|
| 731 |
+
"eval_loss": 2.3006343841552734,
|
| 732 |
+
"eval_runtime": 241.6485,
|
| 733 |
+
"eval_samples_per_second": 827.648,
|
| 734 |
+
"eval_steps_per_second": 51.728,
|
| 735 |
+
"step": 528000
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.26,
|
| 739 |
+
"eval_loss": 2.2969799041748047,
|
| 740 |
+
"eval_runtime": 240.3777,
|
| 741 |
+
"eval_samples_per_second": 832.024,
|
| 742 |
+
"eval_steps_per_second": 52.001,
|
| 743 |
+
"step": 536000
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 2.29,
|
| 747 |
+
"learning_rate": 3.1706666666666665e-07,
|
| 748 |
+
"loss": 2.4454,
|
| 749 |
+
"step": 544000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 2.29,
|
| 753 |
+
"eval_loss": 2.30135178565979,
|
| 754 |
+
"eval_runtime": 241.5787,
|
| 755 |
+
"eval_samples_per_second": 827.887,
|
| 756 |
+
"eval_steps_per_second": 51.743,
|
| 757 |
+
"step": 544000
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 2.32,
|
| 761 |
+
"eval_loss": 2.2992091178894043,
|
| 762 |
+
"eval_runtime": 241.4245,
|
| 763 |
+
"eval_samples_per_second": 828.416,
|
| 764 |
+
"eval_steps_per_second": 51.776,
|
| 765 |
+
"step": 552000
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.36,
|
| 769 |
+
"learning_rate": 3.1433333333333336e-07,
|
| 770 |
+
"loss": 2.4457,
|
| 771 |
+
"step": 560000
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 2.36,
|
| 775 |
+
"eval_loss": 2.2961883544921875,
|
| 776 |
+
"eval_runtime": 242.857,
|
| 777 |
+
"eval_samples_per_second": 823.53,
|
| 778 |
+
"eval_steps_per_second": 51.471,
|
| 779 |
+
"step": 560000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 2.39,
|
| 783 |
+
"eval_loss": 2.300924301147461,
|
| 784 |
+
"eval_runtime": 253.3263,
|
| 785 |
+
"eval_samples_per_second": 789.495,
|
| 786 |
+
"eval_steps_per_second": 49.343,
|
| 787 |
+
"step": 568000
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 2.43,
|
| 791 |
+
"learning_rate": 3.116e-07,
|
| 792 |
+
"loss": 2.4354,
|
| 793 |
+
"step": 576000
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 2.43,
|
| 797 |
+
"eval_loss": 2.2960214614868164,
|
| 798 |
+
"eval_runtime": 251.8236,
|
| 799 |
+
"eval_samples_per_second": 794.207,
|
| 800 |
+
"eval_steps_per_second": 49.638,
|
| 801 |
+
"step": 576000
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.46,
|
| 805 |
+
"eval_loss": 2.3008430004119873,
|
| 806 |
+
"eval_runtime": 255.1,
|
| 807 |
+
"eval_samples_per_second": 784.006,
|
| 808 |
+
"eval_steps_per_second": 49.0,
|
| 809 |
+
"step": 584000
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 2.49,
|
| 813 |
+
"learning_rate": 3.0886666666666667e-07,
|
| 814 |
+
"loss": 2.4361,
|
| 815 |
+
"step": 592000
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 2.49,
|
| 819 |
+
"eval_loss": 2.289783239364624,
|
| 820 |
+
"eval_runtime": 253.1447,
|
| 821 |
+
"eval_samples_per_second": 790.062,
|
| 822 |
+
"eval_steps_per_second": 49.379,
|
| 823 |
+
"step": 592000
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 2.53,
|
| 827 |
+
"eval_loss": 2.306014060974121,
|
| 828 |
+
"eval_runtime": 253.3664,
|
| 829 |
+
"eval_samples_per_second": 789.371,
|
| 830 |
+
"eval_steps_per_second": 49.336,
|
| 831 |
+
"step": 600000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 2.56,
|
| 835 |
+
"learning_rate": 3.061333333333333e-07,
|
| 836 |
+
"loss": 2.4377,
|
| 837 |
+
"step": 608000
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.56,
|
| 841 |
+
"eval_loss": 2.298966884613037,
|
| 842 |
+
"eval_runtime": 254.531,
|
| 843 |
+
"eval_samples_per_second": 785.759,
|
| 844 |
+
"eval_steps_per_second": 49.11,
|
| 845 |
+
"step": 608000
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 2.59,
|
| 849 |
+
"eval_loss": 2.298924684524536,
|
| 850 |
+
"eval_runtime": 253.6979,
|
| 851 |
+
"eval_samples_per_second": 788.339,
|
| 852 |
+
"eval_steps_per_second": 49.271,
|
| 853 |
+
"step": 616000
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 2.63,
|
| 857 |
+
"learning_rate": 3.034e-07,
|
| 858 |
+
"loss": 2.4416,
|
| 859 |
+
"step": 624000
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 2.63,
|
| 863 |
+
"eval_loss": 2.296922206878662,
|
| 864 |
+
"eval_runtime": 252.7348,
|
| 865 |
+
"eval_samples_per_second": 791.343,
|
| 866 |
+
"eval_steps_per_second": 49.459,
|
| 867 |
+
"step": 624000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.66,
|
| 871 |
+
"eval_loss": 2.2932701110839844,
|
| 872 |
+
"eval_runtime": 249.9057,
|
| 873 |
+
"eval_samples_per_second": 800.302,
|
| 874 |
+
"eval_steps_per_second": 50.019,
|
| 875 |
+
"step": 632000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 2.69,
|
| 879 |
+
"learning_rate": 3.0066666666666663e-07,
|
| 880 |
+
"loss": 2.434,
|
| 881 |
+
"step": 640000
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.69,
|
| 885 |
+
"eval_loss": 2.29982852935791,
|
| 886 |
+
"eval_runtime": 250.7868,
|
| 887 |
+
"eval_samples_per_second": 797.49,
|
| 888 |
+
"eval_steps_per_second": 49.843,
|
| 889 |
+
"step": 640000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 2.73,
|
| 893 |
+
"eval_loss": 2.294802665710449,
|
| 894 |
+
"eval_runtime": 251.6635,
|
| 895 |
+
"eval_samples_per_second": 794.712,
|
| 896 |
+
"eval_steps_per_second": 49.669,
|
| 897 |
+
"step": 648000
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 2.76,
|
| 901 |
+
"learning_rate": 2.9793333333333334e-07,
|
| 902 |
+
"loss": 2.433,
|
| 903 |
+
"step": 656000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 2.76,
|
| 907 |
+
"eval_loss": 2.2891652584075928,
|
| 908 |
+
"eval_runtime": 250.4466,
|
| 909 |
+
"eval_samples_per_second": 798.574,
|
| 910 |
+
"eval_steps_per_second": 49.911,
|
| 911 |
+
"step": 656000
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 2.8,
|
| 915 |
+
"eval_loss": 2.292858362197876,
|
| 916 |
+
"eval_runtime": 254.5309,
|
| 917 |
+
"eval_samples_per_second": 785.759,
|
| 918 |
+
"eval_steps_per_second": 49.11,
|
| 919 |
+
"step": 664000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 2.83,
|
| 923 |
+
"learning_rate": 2.952e-07,
|
| 924 |
+
"loss": 2.44,
|
| 925 |
+
"step": 672000
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 2.83,
|
| 929 |
+
"eval_loss": 2.2973103523254395,
|
| 930 |
+
"eval_runtime": 250.641,
|
| 931 |
+
"eval_samples_per_second": 797.954,
|
| 932 |
+
"eval_steps_per_second": 49.872,
|
| 933 |
+
"step": 672000
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 2.86,
|
| 937 |
+
"eval_loss": 2.292598009109497,
|
| 938 |
+
"eval_runtime": 250.7654,
|
| 939 |
+
"eval_samples_per_second": 797.558,
|
| 940 |
+
"eval_steps_per_second": 49.847,
|
| 941 |
+
"step": 680000
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.9,
|
| 945 |
+
"learning_rate": 2.9246666666666665e-07,
|
| 946 |
+
"loss": 2.4291,
|
| 947 |
+
"step": 688000
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 2.9,
|
| 951 |
+
"eval_loss": 2.298957109451294,
|
| 952 |
+
"eval_runtime": 249.6027,
|
| 953 |
+
"eval_samples_per_second": 801.273,
|
| 954 |
+
"eval_steps_per_second": 50.08,
|
| 955 |
+
"step": 688000
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.93,
|
| 959 |
+
"eval_loss": 2.2936558723449707,
|
| 960 |
+
"eval_runtime": 254.0075,
|
| 961 |
+
"eval_samples_per_second": 787.378,
|
| 962 |
+
"eval_steps_per_second": 49.211,
|
| 963 |
+
"step": 696000
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.96,
|
| 967 |
+
"learning_rate": 2.897333333333333e-07,
|
| 968 |
+
"loss": 2.4336,
|
| 969 |
+
"step": 704000
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 2.96,
|
| 973 |
+
"eval_loss": 2.2894132137298584,
|
| 974 |
+
"eval_runtime": 251.6948,
|
| 975 |
+
"eval_samples_per_second": 794.613,
|
| 976 |
+
"eval_steps_per_second": 49.663,
|
| 977 |
+
"step": 704000
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 3.0,
|
| 981 |
+
"eval_loss": 2.295807123184204,
|
| 982 |
+
"eval_runtime": 250.3917,
|
| 983 |
+
"eval_samples_per_second": 798.748,
|
| 984 |
+
"eval_steps_per_second": 49.922,
|
| 985 |
+
"step": 712000
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 3.03,
|
| 989 |
+
"learning_rate": 2.8699999999999996e-07,
|
| 990 |
+
"loss": 2.439,
|
| 991 |
+
"step": 720000
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 3.03,
|
| 995 |
+
"eval_loss": 2.295633554458618,
|
| 996 |
+
"eval_runtime": 249.5836,
|
| 997 |
+
"eval_samples_per_second": 801.335,
|
| 998 |
+
"eval_steps_per_second": 50.083,
|
| 999 |
+
"step": 720000
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 3.07,
|
| 1003 |
+
"eval_loss": 2.2927510738372803,
|
| 1004 |
+
"eval_runtime": 252.6433,
|
| 1005 |
+
"eval_samples_per_second": 791.63,
|
| 1006 |
+
"eval_steps_per_second": 49.477,
|
| 1007 |
+
"step": 728000
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 3.1,
|
| 1011 |
+
"learning_rate": 2.8426666666666667e-07,
|
| 1012 |
+
"loss": 2.4405,
|
| 1013 |
+
"step": 736000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 3.1,
|
| 1017 |
+
"eval_loss": 2.295588970184326,
|
| 1018 |
+
"eval_runtime": 254.3928,
|
| 1019 |
+
"eval_samples_per_second": 786.186,
|
| 1020 |
+
"eval_steps_per_second": 49.137,
|
| 1021 |
+
"step": 736000
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 3.13,
|
| 1025 |
+
"eval_loss": 2.290548324584961,
|
| 1026 |
+
"eval_runtime": 251.5538,
|
| 1027 |
+
"eval_samples_per_second": 795.058,
|
| 1028 |
+
"eval_steps_per_second": 49.691,
|
| 1029 |
+
"step": 744000
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 3.17,
|
| 1033 |
+
"learning_rate": 2.815333333333333e-07,
|
| 1034 |
+
"loss": 2.4332,
|
| 1035 |
+
"step": 752000
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 3.17,
|
| 1039 |
+
"eval_loss": 2.2921102046966553,
|
| 1040 |
+
"eval_runtime": 250.7706,
|
| 1041 |
+
"eval_samples_per_second": 797.542,
|
| 1042 |
+
"eval_steps_per_second": 49.846,
|
| 1043 |
+
"step": 752000
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 3.2,
|
| 1047 |
+
"eval_loss": 2.2907326221466064,
|
| 1048 |
+
"eval_runtime": 252.731,
|
| 1049 |
+
"eval_samples_per_second": 791.355,
|
| 1050 |
+
"eval_steps_per_second": 49.46,
|
| 1051 |
+
"step": 760000
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 3.23,
|
| 1055 |
+
"learning_rate": 2.7880000000000003e-07,
|
| 1056 |
+
"loss": 2.4353,
|
| 1057 |
+
"step": 768000
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 3.23,
|
| 1061 |
+
"eval_loss": 2.2879087924957275,
|
| 1062 |
+
"eval_runtime": 249.8593,
|
| 1063 |
+
"eval_samples_per_second": 800.45,
|
| 1064 |
+
"eval_steps_per_second": 50.028,
|
| 1065 |
+
"step": 768000
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 3.27,
|
| 1069 |
+
"eval_loss": 2.292853355407715,
|
| 1070 |
+
"eval_runtime": 255.5473,
|
| 1071 |
+
"eval_samples_per_second": 782.634,
|
| 1072 |
+
"eval_steps_per_second": 48.915,
|
| 1073 |
+
"step": 776000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 3.3,
|
| 1077 |
+
"learning_rate": 2.7606666666666664e-07,
|
| 1078 |
+
"loss": 2.4273,
|
| 1079 |
+
"step": 784000
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 3.3,
|
| 1083 |
+
"eval_loss": 2.29532527923584,
|
| 1084 |
+
"eval_runtime": 249.121,
|
| 1085 |
+
"eval_samples_per_second": 802.823,
|
| 1086 |
+
"eval_steps_per_second": 50.176,
|
| 1087 |
+
"step": 784000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 3.33,
|
| 1091 |
+
"eval_loss": 2.2917468547821045,
|
| 1092 |
+
"eval_runtime": 252.7811,
|
| 1093 |
+
"eval_samples_per_second": 791.198,
|
| 1094 |
+
"eval_steps_per_second": 49.45,
|
| 1095 |
+
"step": 792000
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 3.37,
|
| 1099 |
+
"learning_rate": 2.733333333333333e-07,
|
| 1100 |
+
"loss": 2.4233,
|
| 1101 |
+
"step": 800000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 3.37,
|
| 1105 |
+
"eval_loss": 2.294677734375,
|
| 1106 |
+
"eval_runtime": 251.4379,
|
| 1107 |
+
"eval_samples_per_second": 795.425,
|
| 1108 |
+
"eval_steps_per_second": 49.714,
|
| 1109 |
+
"step": 800000
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 3.4,
|
| 1113 |
+
"eval_loss": 2.294262170791626,
|
| 1114 |
+
"eval_runtime": 249.692,
|
| 1115 |
+
"eval_samples_per_second": 800.987,
|
| 1116 |
+
"eval_steps_per_second": 50.062,
|
| 1117 |
+
"step": 808000
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 3.44,
|
| 1121 |
+
"learning_rate": 2.706e-07,
|
| 1122 |
+
"loss": 2.4324,
|
| 1123 |
+
"step": 816000
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 3.44,
|
| 1127 |
+
"eval_loss": 2.2940258979797363,
|
| 1128 |
+
"eval_runtime": 251.4645,
|
| 1129 |
+
"eval_samples_per_second": 795.341,
|
| 1130 |
+
"eval_steps_per_second": 49.709,
|
| 1131 |
+
"step": 816000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 3.47,
|
| 1135 |
+
"eval_loss": 2.291130781173706,
|
| 1136 |
+
"eval_runtime": 250.9475,
|
| 1137 |
+
"eval_samples_per_second": 796.979,
|
| 1138 |
+
"eval_steps_per_second": 49.811,
|
| 1139 |
+
"step": 824000
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 3.5,
|
| 1143 |
+
"learning_rate": 2.6786666666666666e-07,
|
| 1144 |
+
"loss": 2.4461,
|
| 1145 |
+
"step": 832000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 3.5,
|
| 1149 |
+
"eval_loss": 2.2919955253601074,
|
| 1150 |
+
"eval_runtime": 251.5475,
|
| 1151 |
+
"eval_samples_per_second": 795.078,
|
| 1152 |
+
"eval_steps_per_second": 49.692,
|
| 1153 |
+
"step": 832000
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 3.54,
|
| 1157 |
+
"eval_loss": 2.291116952896118,
|
| 1158 |
+
"eval_runtime": 247.564,
|
| 1159 |
+
"eval_samples_per_second": 807.872,
|
| 1160 |
+
"eval_steps_per_second": 50.492,
|
| 1161 |
+
"step": 840000
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 3.57,
|
| 1165 |
+
"learning_rate": 2.651333333333333e-07,
|
| 1166 |
+
"loss": 2.4267,
|
| 1167 |
+
"step": 848000
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 3.57,
|
| 1171 |
+
"eval_loss": 2.294036865234375,
|
| 1172 |
+
"eval_runtime": 248.8076,
|
| 1173 |
+
"eval_samples_per_second": 803.834,
|
| 1174 |
+
"eval_steps_per_second": 50.24,
|
| 1175 |
+
"step": 848000
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 3.6,
|
| 1179 |
+
"eval_loss": 2.2889564037323,
|
| 1180 |
+
"eval_runtime": 250.7173,
|
| 1181 |
+
"eval_samples_per_second": 797.711,
|
| 1182 |
+
"eval_steps_per_second": 49.857,
|
| 1183 |
+
"step": 856000
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 3.64,
|
| 1187 |
+
"learning_rate": 2.624e-07,
|
| 1188 |
+
"loss": 2.4313,
|
| 1189 |
+
"step": 864000
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 3.64,
|
| 1193 |
+
"eval_loss": 2.2913272380828857,
|
| 1194 |
+
"eval_runtime": 251.8602,
|
| 1195 |
+
"eval_samples_per_second": 794.091,
|
| 1196 |
+
"eval_steps_per_second": 49.631,
|
| 1197 |
+
"step": 864000
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 3.67,
|
| 1201 |
+
"eval_loss": 2.296712875366211,
|
| 1202 |
+
"eval_runtime": 250.3469,
|
| 1203 |
+
"eval_samples_per_second": 798.891,
|
| 1204 |
+
"eval_steps_per_second": 49.931,
|
| 1205 |
+
"step": 872000
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 3.71,
|
| 1209 |
+
"learning_rate": 2.596666666666667e-07,
|
| 1210 |
+
"loss": 2.4388,
|
| 1211 |
+
"step": 880000
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 3.71,
|
| 1215 |
+
"eval_loss": 2.29068660736084,
|
| 1216 |
+
"eval_runtime": 252.8984,
|
| 1217 |
+
"eval_samples_per_second": 790.831,
|
| 1218 |
+
"eval_steps_per_second": 49.427,
|
| 1219 |
+
"step": 880000
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 3.74,
|
| 1223 |
+
"eval_loss": 2.295198440551758,
|
| 1224 |
+
"eval_runtime": 250.9204,
|
| 1225 |
+
"eval_samples_per_second": 797.065,
|
| 1226 |
+
"eval_steps_per_second": 49.817,
|
| 1227 |
+
"step": 888000
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 3.77,
|
| 1231 |
+
"learning_rate": 2.5693333333333333e-07,
|
| 1232 |
+
"loss": 2.4326,
|
| 1233 |
+
"step": 896000
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 3.77,
|
| 1237 |
+
"eval_loss": 2.287299394607544,
|
| 1238 |
+
"eval_runtime": 250.3015,
|
| 1239 |
+
"eval_samples_per_second": 799.036,
|
| 1240 |
+
"eval_steps_per_second": 49.94,
|
| 1241 |
+
"step": 896000
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 3.81,
|
| 1245 |
+
"eval_loss": 2.287095308303833,
|
| 1246 |
+
"eval_runtime": 251.9856,
|
| 1247 |
+
"eval_samples_per_second": 793.696,
|
| 1248 |
+
"eval_steps_per_second": 49.606,
|
| 1249 |
+
"step": 904000
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 3.84,
|
| 1253 |
+
"learning_rate": 2.542e-07,
|
| 1254 |
+
"loss": 2.4312,
|
| 1255 |
+
"step": 912000
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 3.84,
|
| 1259 |
+
"eval_loss": 2.2880072593688965,
|
| 1260 |
+
"eval_runtime": 250.0793,
|
| 1261 |
+
"eval_samples_per_second": 799.746,
|
| 1262 |
+
"eval_steps_per_second": 49.984,
|
| 1263 |
+
"step": 912000
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 3.87,
|
| 1267 |
+
"eval_loss": 2.2941174507141113,
|
| 1268 |
+
"eval_runtime": 250.5896,
|
| 1269 |
+
"eval_samples_per_second": 798.118,
|
| 1270 |
+
"eval_steps_per_second": 49.882,
|
| 1271 |
+
"step": 920000
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 3.91,
|
| 1275 |
+
"learning_rate": 2.5146666666666664e-07,
|
| 1276 |
+
"loss": 2.4398,
|
| 1277 |
+
"step": 928000
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 3.91,
|
| 1281 |
+
"eval_loss": 2.292538642883301,
|
| 1282 |
+
"eval_runtime": 248.6711,
|
| 1283 |
+
"eval_samples_per_second": 804.275,
|
| 1284 |
+
"eval_steps_per_second": 50.267,
|
| 1285 |
+
"step": 928000
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 3.94,
|
| 1289 |
+
"eval_loss": 2.296539545059204,
|
| 1290 |
+
"eval_runtime": 250.5232,
|
| 1291 |
+
"eval_samples_per_second": 798.329,
|
| 1292 |
+
"eval_steps_per_second": 49.896,
|
| 1293 |
+
"step": 936000
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 3.97,
|
| 1297 |
+
"learning_rate": 2.4873333333333335e-07,
|
| 1298 |
+
"loss": 2.441,
|
| 1299 |
+
"step": 944000
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 3.97,
|
| 1303 |
+
"eval_loss": 2.289044141769409,
|
| 1304 |
+
"eval_runtime": 251.7273,
|
| 1305 |
+
"eval_samples_per_second": 794.511,
|
| 1306 |
+
"eval_steps_per_second": 49.657,
|
| 1307 |
+
"step": 944000
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 4.01,
|
| 1311 |
+
"eval_loss": 2.2945752143859863,
|
| 1312 |
+
"eval_runtime": 250.3966,
|
| 1313 |
+
"eval_samples_per_second": 798.733,
|
| 1314 |
+
"eval_steps_per_second": 49.921,
|
| 1315 |
+
"step": 952000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 4.04,
|
| 1319 |
+
"learning_rate": 2.46e-07,
|
| 1320 |
+
"loss": 2.4345,
|
| 1321 |
+
"step": 960000
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 4.04,
|
| 1325 |
+
"eval_loss": 2.291003704071045,
|
| 1326 |
+
"eval_runtime": 250.1466,
|
| 1327 |
+
"eval_samples_per_second": 799.531,
|
| 1328 |
+
"eval_steps_per_second": 49.971,
|
| 1329 |
+
"step": 960000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 4.08,
|
| 1333 |
+
"eval_loss": 2.279160737991333,
|
| 1334 |
+
"eval_runtime": 251.3138,
|
| 1335 |
+
"eval_samples_per_second": 795.818,
|
| 1336 |
+
"eval_steps_per_second": 49.739,
|
| 1337 |
+
"step": 968000
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 4.11,
|
| 1341 |
+
"learning_rate": 2.4326666666666666e-07,
|
| 1342 |
+
"loss": 2.4332,
|
| 1343 |
+
"step": 976000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 4.11,
|
| 1347 |
+
"eval_loss": 2.285647392272949,
|
| 1348 |
+
"eval_runtime": 251.6666,
|
| 1349 |
+
"eval_samples_per_second": 794.702,
|
| 1350 |
+
"eval_steps_per_second": 49.669,
|
| 1351 |
+
"step": 976000
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 4.14,
|
| 1355 |
+
"eval_loss": 2.2878894805908203,
|
| 1356 |
+
"eval_runtime": 249.5029,
|
| 1357 |
+
"eval_samples_per_second": 801.594,
|
| 1358 |
+
"eval_steps_per_second": 50.1,
|
| 1359 |
+
"step": 984000
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 4.18,
|
| 1363 |
+
"learning_rate": 2.405333333333333e-07,
|
| 1364 |
+
"loss": 2.4375,
|
| 1365 |
+
"step": 992000
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 4.18,
|
| 1369 |
+
"eval_loss": 2.2861104011535645,
|
| 1370 |
+
"eval_runtime": 249.5689,
|
| 1371 |
+
"eval_samples_per_second": 801.382,
|
| 1372 |
+
"eval_steps_per_second": 50.086,
|
| 1373 |
+
"step": 992000
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 4.21,
|
| 1377 |
+
"eval_loss": 2.2892367839813232,
|
| 1378 |
+
"eval_runtime": 250.1781,
|
| 1379 |
+
"eval_samples_per_second": 799.43,
|
| 1380 |
+
"eval_steps_per_second": 49.964,
|
| 1381 |
+
"step": 1000000
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 4.24,
|
| 1385 |
+
"learning_rate": 2.3779999999999997e-07,
|
| 1386 |
+
"loss": 2.4282,
|
| 1387 |
+
"step": 1008000
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 4.24,
|
| 1391 |
+
"eval_loss": 2.2894413471221924,
|
| 1392 |
+
"eval_runtime": 252.3508,
|
| 1393 |
+
"eval_samples_per_second": 792.548,
|
| 1394 |
+
"eval_steps_per_second": 49.534,
|
| 1395 |
+
"step": 1008000
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 4.28,
|
| 1399 |
+
"eval_loss": 2.290153980255127,
|
| 1400 |
+
"eval_runtime": 250.4413,
|
| 1401 |
+
"eval_samples_per_second": 798.59,
|
| 1402 |
+
"eval_steps_per_second": 49.912,
|
| 1403 |
+
"step": 1016000
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 4.31,
|
| 1407 |
+
"learning_rate": 2.3506666666666668e-07,
|
| 1408 |
+
"loss": 2.4231,
|
| 1409 |
+
"step": 1024000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 4.31,
|
| 1413 |
+
"eval_loss": 2.282951831817627,
|
| 1414 |
+
"eval_runtime": 251.0777,
|
| 1415 |
+
"eval_samples_per_second": 796.566,
|
| 1416 |
+
"eval_steps_per_second": 49.785,
|
| 1417 |
+
"step": 1024000
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 4.35,
|
| 1421 |
+
"eval_loss": 2.294783353805542,
|
| 1422 |
+
"eval_runtime": 249.9563,
|
| 1423 |
+
"eval_samples_per_second": 800.14,
|
| 1424 |
+
"eval_steps_per_second": 50.009,
|
| 1425 |
+
"step": 1032000
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 4.38,
|
| 1429 |
+
"learning_rate": 2.3233333333333334e-07,
|
| 1430 |
+
"loss": 2.4299,
|
| 1431 |
+
"step": 1040000
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 4.38,
|
| 1435 |
+
"eval_loss": 2.2915213108062744,
|
| 1436 |
+
"eval_runtime": 251.5141,
|
| 1437 |
+
"eval_samples_per_second": 795.184,
|
| 1438 |
+
"eval_steps_per_second": 49.699,
|
| 1439 |
+
"step": 1040000
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 4.41,
|
| 1443 |
+
"eval_loss": 2.2921574115753174,
|
| 1444 |
+
"eval_runtime": 250.6578,
|
| 1445 |
+
"eval_samples_per_second": 797.9,
|
| 1446 |
+
"eval_steps_per_second": 49.869,
|
| 1447 |
+
"step": 1048000
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 4.45,
|
| 1451 |
+
"learning_rate": 2.2960000000000002e-07,
|
| 1452 |
+
"loss": 2.4353,
|
| 1453 |
+
"step": 1056000
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 4.45,
|
| 1457 |
+
"eval_loss": 2.287623882293701,
|
| 1458 |
+
"eval_runtime": 249.9183,
|
| 1459 |
+
"eval_samples_per_second": 800.262,
|
| 1460 |
+
"eval_steps_per_second": 50.016,
|
| 1461 |
+
"step": 1056000
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 4.48,
|
| 1465 |
+
"eval_loss": 2.2892725467681885,
|
| 1466 |
+
"eval_runtime": 249.4102,
|
| 1467 |
+
"eval_samples_per_second": 801.892,
|
| 1468 |
+
"eval_steps_per_second": 50.118,
|
| 1469 |
+
"step": 1064000
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 4.51,
|
| 1473 |
+
"learning_rate": 2.2686666666666667e-07,
|
| 1474 |
+
"loss": 2.4308,
|
| 1475 |
+
"step": 1072000
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 4.51,
|
| 1479 |
+
"eval_loss": 2.292046070098877,
|
| 1480 |
+
"eval_runtime": 247.5504,
|
| 1481 |
+
"eval_samples_per_second": 807.916,
|
| 1482 |
+
"eval_steps_per_second": 50.495,
|
| 1483 |
+
"step": 1072000
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 4.55,
|
| 1487 |
+
"eval_loss": 2.285968780517578,
|
| 1488 |
+
"eval_runtime": 247.5335,
|
| 1489 |
+
"eval_samples_per_second": 807.971,
|
| 1490 |
+
"eval_steps_per_second": 50.498,
|
| 1491 |
+
"step": 1080000
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 4.58,
|
| 1495 |
+
"learning_rate": 2.2413333333333333e-07,
|
| 1496 |
+
"loss": 2.4358,
|
| 1497 |
+
"step": 1088000
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 4.58,
|
| 1501 |
+
"eval_loss": 2.2907297611236572,
|
| 1502 |
+
"eval_runtime": 248.9275,
|
| 1503 |
+
"eval_samples_per_second": 803.447,
|
| 1504 |
+
"eval_steps_per_second": 50.215,
|
| 1505 |
+
"step": 1088000
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 4.61,
|
| 1509 |
+
"eval_loss": 2.2807979583740234,
|
| 1510 |
+
"eval_runtime": 249.3116,
|
| 1511 |
+
"eval_samples_per_second": 802.209,
|
| 1512 |
+
"eval_steps_per_second": 50.138,
|
| 1513 |
+
"step": 1096000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 4.65,
|
| 1517 |
+
"learning_rate": 2.214e-07,
|
| 1518 |
+
"loss": 2.4341,
|
| 1519 |
+
"step": 1104000
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 4.65,
|
| 1523 |
+
"eval_loss": 2.290224313735962,
|
| 1524 |
+
"eval_runtime": 249.5198,
|
| 1525 |
+
"eval_samples_per_second": 801.54,
|
| 1526 |
+
"eval_steps_per_second": 50.096,
|
| 1527 |
+
"step": 1104000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 4.68,
|
| 1531 |
+
"eval_loss": 2.281525135040283,
|
| 1532 |
+
"eval_runtime": 249.0717,
|
| 1533 |
+
"eval_samples_per_second": 802.982,
|
| 1534 |
+
"eval_steps_per_second": 50.186,
|
| 1535 |
+
"step": 1112000
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 4.72,
|
| 1539 |
+
"learning_rate": 2.1866666666666667e-07,
|
| 1540 |
+
"loss": 2.4315,
|
| 1541 |
+
"step": 1120000
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 4.72,
|
| 1545 |
+
"eval_loss": 2.296067237854004,
|
| 1546 |
+
"eval_runtime": 246.4788,
|
| 1547 |
+
"eval_samples_per_second": 811.429,
|
| 1548 |
+
"eval_steps_per_second": 50.714,
|
| 1549 |
+
"step": 1120000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 4.75,
|
| 1553 |
+
"eval_loss": 2.288472890853882,
|
| 1554 |
+
"eval_runtime": 243.6257,
|
| 1555 |
+
"eval_samples_per_second": 820.932,
|
| 1556 |
+
"eval_steps_per_second": 51.308,
|
| 1557 |
+
"step": 1128000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 4.78,
|
| 1561 |
+
"learning_rate": 2.1593333333333332e-07,
|
| 1562 |
+
"loss": 2.434,
|
| 1563 |
+
"step": 1136000
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 4.78,
|
| 1567 |
+
"eval_loss": 2.2916722297668457,
|
| 1568 |
+
"eval_runtime": 242.6981,
|
| 1569 |
+
"eval_samples_per_second": 824.069,
|
| 1570 |
+
"eval_steps_per_second": 51.504,
|
| 1571 |
+
"step": 1136000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 4.82,
|
| 1575 |
+
"eval_loss": 2.285139560699463,
|
| 1576 |
+
"eval_runtime": 243.3147,
|
| 1577 |
+
"eval_samples_per_second": 821.981,
|
| 1578 |
+
"eval_steps_per_second": 51.374,
|
| 1579 |
+
"step": 1144000
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 4.85,
|
| 1583 |
+
"learning_rate": 2.132e-07,
|
| 1584 |
+
"loss": 2.4324,
|
| 1585 |
+
"step": 1152000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 4.85,
|
| 1589 |
+
"eval_loss": 2.2837140560150146,
|
| 1590 |
+
"eval_runtime": 243.7667,
|
| 1591 |
+
"eval_samples_per_second": 820.457,
|
| 1592 |
+
"eval_steps_per_second": 51.279,
|
| 1593 |
+
"step": 1152000
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 4.88,
|
| 1597 |
+
"eval_loss": 2.288276195526123,
|
| 1598 |
+
"eval_runtime": 243.0553,
|
| 1599 |
+
"eval_samples_per_second": 822.858,
|
| 1600 |
+
"eval_steps_per_second": 51.429,
|
| 1601 |
+
"step": 1160000
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 4.92,
|
| 1605 |
+
"learning_rate": 2.1046666666666666e-07,
|
| 1606 |
+
"loss": 2.4297,
|
| 1607 |
+
"step": 1168000
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 4.92,
|
| 1611 |
+
"eval_loss": 2.282362222671509,
|
| 1612 |
+
"eval_runtime": 243.4119,
|
| 1613 |
+
"eval_samples_per_second": 821.652,
|
| 1614 |
+
"eval_steps_per_second": 51.353,
|
| 1615 |
+
"step": 1168000
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 4.95,
|
| 1619 |
+
"eval_loss": 2.283245086669922,
|
| 1620 |
+
"eval_runtime": 242.8004,
|
| 1621 |
+
"eval_samples_per_second": 823.722,
|
| 1622 |
+
"eval_steps_per_second": 51.483,
|
| 1623 |
+
"step": 1176000
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 4.99,
|
| 1627 |
+
"learning_rate": 2.0773333333333334e-07,
|
| 1628 |
+
"loss": 2.436,
|
| 1629 |
+
"step": 1184000
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 4.99,
|
| 1633 |
+
"eval_loss": 2.286545515060425,
|
| 1634 |
+
"eval_runtime": 243.7626,
|
| 1635 |
+
"eval_samples_per_second": 820.471,
|
| 1636 |
+
"eval_steps_per_second": 51.279,
|
| 1637 |
+
"step": 1184000
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 5.02,
|
| 1641 |
+
"eval_loss": 2.281639337539673,
|
| 1642 |
+
"eval_runtime": 242.7206,
|
| 1643 |
+
"eval_samples_per_second": 823.993,
|
| 1644 |
+
"eval_steps_per_second": 51.5,
|
| 1645 |
+
"step": 1192000
|
| 1646 |
+
},
|
| 1647 |
+
{
|
| 1648 |
+
"epoch": 5.05,
|
| 1649 |
+
"learning_rate": 2.05e-07,
|
| 1650 |
+
"loss": 2.4329,
|
| 1651 |
+
"step": 1200000
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 5.05,
|
| 1655 |
+
"eval_loss": 2.2862467765808105,
|
| 1656 |
+
"eval_runtime": 242.9508,
|
| 1657 |
+
"eval_samples_per_second": 823.212,
|
| 1658 |
+
"eval_steps_per_second": 51.451,
|
| 1659 |
+
"step": 1200000
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 5.09,
|
| 1663 |
+
"eval_loss": 2.284660577774048,
|
| 1664 |
+
"eval_runtime": 244.035,
|
| 1665 |
+
"eval_samples_per_second": 819.554,
|
| 1666 |
+
"eval_steps_per_second": 51.222,
|
| 1667 |
+
"step": 1208000
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 5.12,
|
| 1671 |
+
"learning_rate": 2.0226666666666668e-07,
|
| 1672 |
+
"loss": 2.4276,
|
| 1673 |
+
"step": 1216000
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 5.12,
|
| 1677 |
+
"eval_loss": 2.295132875442505,
|
| 1678 |
+
"eval_runtime": 243.547,
|
| 1679 |
+
"eval_samples_per_second": 821.197,
|
| 1680 |
+
"eval_steps_per_second": 51.325,
|
| 1681 |
+
"step": 1216000
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 5.15,
|
| 1685 |
+
"eval_loss": 2.297971487045288,
|
| 1686 |
+
"eval_runtime": 243.0459,
|
| 1687 |
+
"eval_samples_per_second": 822.89,
|
| 1688 |
+
"eval_steps_per_second": 51.431,
|
| 1689 |
+
"step": 1224000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 5.19,
|
| 1693 |
+
"learning_rate": 1.9953333333333333e-07,
|
| 1694 |
+
"loss": 2.4362,
|
| 1695 |
+
"step": 1232000
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 5.19,
|
| 1699 |
+
"eval_loss": 2.2889420986175537,
|
| 1700 |
+
"eval_runtime": 245.3823,
|
| 1701 |
+
"eval_samples_per_second": 815.055,
|
| 1702 |
+
"eval_steps_per_second": 50.941,
|
| 1703 |
+
"step": 1232000
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 5.22,
|
| 1707 |
+
"eval_loss": 2.2913663387298584,
|
| 1708 |
+
"eval_runtime": 243.0295,
|
| 1709 |
+
"eval_samples_per_second": 822.945,
|
| 1710 |
+
"eval_steps_per_second": 51.434,
|
| 1711 |
+
"step": 1240000
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 5.25,
|
| 1715 |
+
"learning_rate": 1.968e-07,
|
| 1716 |
+
"loss": 2.4309,
|
| 1717 |
+
"step": 1248000
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 5.25,
|
| 1721 |
+
"eval_loss": 2.2914836406707764,
|
| 1722 |
+
"eval_runtime": 245.1791,
|
| 1723 |
+
"eval_samples_per_second": 815.73,
|
| 1724 |
+
"eval_steps_per_second": 50.983,
|
| 1725 |
+
"step": 1248000
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 5.29,
|
| 1729 |
+
"eval_loss": 2.2822399139404297,
|
| 1730 |
+
"eval_runtime": 243.1242,
|
| 1731 |
+
"eval_samples_per_second": 822.625,
|
| 1732 |
+
"eval_steps_per_second": 51.414,
|
| 1733 |
+
"step": 1256000
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 5.32,
|
| 1737 |
+
"learning_rate": 1.9406666666666667e-07,
|
| 1738 |
+
"loss": 2.4414,
|
| 1739 |
+
"step": 1264000
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 5.32,
|
| 1743 |
+
"eval_loss": 2.287137269973755,
|
| 1744 |
+
"eval_runtime": 243.2167,
|
| 1745 |
+
"eval_samples_per_second": 822.312,
|
| 1746 |
+
"eval_steps_per_second": 51.395,
|
| 1747 |
+
"step": 1264000
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 5.36,
|
| 1751 |
+
"eval_loss": 2.2890405654907227,
|
| 1752 |
+
"eval_runtime": 246.4538,
|
| 1753 |
+
"eval_samples_per_second": 811.511,
|
| 1754 |
+
"eval_steps_per_second": 50.719,
|
| 1755 |
+
"step": 1272000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 5.39,
|
| 1759 |
+
"learning_rate": 1.9133333333333333e-07,
|
| 1760 |
+
"loss": 2.4241,
|
| 1761 |
+
"step": 1280000
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 5.39,
|
| 1765 |
+
"eval_loss": 2.2844035625457764,
|
| 1766 |
+
"eval_runtime": 243.8151,
|
| 1767 |
+
"eval_samples_per_second": 820.294,
|
| 1768 |
+
"eval_steps_per_second": 51.268,
|
| 1769 |
+
"step": 1280000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 5.42,
|
| 1773 |
+
"eval_loss": 2.281158924102783,
|
| 1774 |
+
"eval_runtime": 244.4232,
|
| 1775 |
+
"eval_samples_per_second": 818.253,
|
| 1776 |
+
"eval_steps_per_second": 51.141,
|
| 1777 |
+
"step": 1288000
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 5.46,
|
| 1781 |
+
"learning_rate": 1.886e-07,
|
| 1782 |
+
"loss": 2.4251,
|
| 1783 |
+
"step": 1296000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 5.46,
|
| 1787 |
+
"eval_loss": 2.2873995304107666,
|
| 1788 |
+
"eval_runtime": 245.6666,
|
| 1789 |
+
"eval_samples_per_second": 814.111,
|
| 1790 |
+
"eval_steps_per_second": 50.882,
|
| 1791 |
+
"step": 1296000
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 5.49,
|
| 1795 |
+
"eval_loss": 2.2846264839172363,
|
| 1796 |
+
"eval_runtime": 244.5062,
|
| 1797 |
+
"eval_samples_per_second": 817.975,
|
| 1798 |
+
"eval_steps_per_second": 51.123,
|
| 1799 |
+
"step": 1304000
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 5.52,
|
| 1803 |
+
"learning_rate": 1.8586666666666666e-07,
|
| 1804 |
+
"loss": 2.4318,
|
| 1805 |
+
"step": 1312000
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 5.52,
|
| 1809 |
+
"eval_loss": 2.283116340637207,
|
| 1810 |
+
"eval_runtime": 245.4301,
|
| 1811 |
+
"eval_samples_per_second": 814.896,
|
| 1812 |
+
"eval_steps_per_second": 50.931,
|
| 1813 |
+
"step": 1312000
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 5.56,
|
| 1817 |
+
"eval_loss": 2.2895309925079346,
|
| 1818 |
+
"eval_runtime": 244.3142,
|
| 1819 |
+
"eval_samples_per_second": 818.618,
|
| 1820 |
+
"eval_steps_per_second": 51.164,
|
| 1821 |
+
"step": 1320000
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 5.59,
|
| 1825 |
+
"learning_rate": 1.8313333333333332e-07,
|
| 1826 |
+
"loss": 2.4247,
|
| 1827 |
+
"step": 1328000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 5.59,
|
| 1831 |
+
"eval_loss": 2.2795751094818115,
|
| 1832 |
+
"eval_runtime": 246.3547,
|
| 1833 |
+
"eval_samples_per_second": 811.838,
|
| 1834 |
+
"eval_steps_per_second": 50.74,
|
| 1835 |
+
"step": 1328000
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 5.63,
|
| 1839 |
+
"eval_loss": 2.2833728790283203,
|
| 1840 |
+
"eval_runtime": 244.3038,
|
| 1841 |
+
"eval_samples_per_second": 818.653,
|
| 1842 |
+
"eval_steps_per_second": 51.166,
|
| 1843 |
+
"step": 1336000
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 5.66,
|
| 1847 |
+
"learning_rate": 1.804e-07,
|
| 1848 |
+
"loss": 2.4305,
|
| 1849 |
+
"step": 1344000
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 5.66,
|
| 1853 |
+
"eval_loss": 2.2810542583465576,
|
| 1854 |
+
"eval_runtime": 245.8027,
|
| 1855 |
+
"eval_samples_per_second": 813.661,
|
| 1856 |
+
"eval_steps_per_second": 50.854,
|
| 1857 |
+
"step": 1344000
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 5.69,
|
| 1861 |
+
"eval_loss": 2.2922263145446777,
|
| 1862 |
+
"eval_runtime": 244.3162,
|
| 1863 |
+
"eval_samples_per_second": 818.611,
|
| 1864 |
+
"eval_steps_per_second": 51.163,
|
| 1865 |
+
"step": 1352000
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 5.73,
|
| 1869 |
+
"learning_rate": 1.7766666666666666e-07,
|
| 1870 |
+
"loss": 2.4336,
|
| 1871 |
+
"step": 1360000
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 5.73,
|
| 1875 |
+
"eval_loss": 2.2830028533935547,
|
| 1876 |
+
"eval_runtime": 245.6363,
|
| 1877 |
+
"eval_samples_per_second": 814.212,
|
| 1878 |
+
"eval_steps_per_second": 50.888,
|
| 1879 |
+
"step": 1360000
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 5.76,
|
| 1883 |
+
"eval_loss": 2.2903530597686768,
|
| 1884 |
+
"eval_runtime": 244.3808,
|
| 1885 |
+
"eval_samples_per_second": 818.395,
|
| 1886 |
+
"eval_steps_per_second": 51.15,
|
| 1887 |
+
"step": 1368000
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 5.79,
|
| 1891 |
+
"learning_rate": 1.7493333333333334e-07,
|
| 1892 |
+
"loss": 2.428,
|
| 1893 |
+
"step": 1376000
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 5.79,
|
| 1897 |
+
"eval_loss": 2.284269332885742,
|
| 1898 |
+
"eval_runtime": 243.8962,
|
| 1899 |
+
"eval_samples_per_second": 820.021,
|
| 1900 |
+
"eval_steps_per_second": 51.251,
|
| 1901 |
+
"step": 1376000
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 5.83,
|
| 1905 |
+
"eval_loss": 2.2804477214813232,
|
| 1906 |
+
"eval_runtime": 243.7523,
|
| 1907 |
+
"eval_samples_per_second": 820.505,
|
| 1908 |
+
"eval_steps_per_second": 51.282,
|
| 1909 |
+
"step": 1384000
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 5.86,
|
| 1913 |
+
"learning_rate": 1.722e-07,
|
| 1914 |
+
"loss": 2.4254,
|
| 1915 |
+
"step": 1392000
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 5.86,
|
| 1919 |
+
"eval_loss": 2.285198211669922,
|
| 1920 |
+
"eval_runtime": 244.494,
|
| 1921 |
+
"eval_samples_per_second": 818.016,
|
| 1922 |
+
"eval_steps_per_second": 51.126,
|
| 1923 |
+
"step": 1392000
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 5.89,
|
| 1927 |
+
"eval_loss": 2.285767078399658,
|
| 1928 |
+
"eval_runtime": 245.3316,
|
| 1929 |
+
"eval_samples_per_second": 815.223,
|
| 1930 |
+
"eval_steps_per_second": 50.951,
|
| 1931 |
+
"step": 1400000
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 5.93,
|
| 1935 |
+
"learning_rate": 1.6946666666666668e-07,
|
| 1936 |
+
"loss": 2.4287,
|
| 1937 |
+
"step": 1408000
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 5.93,
|
| 1941 |
+
"eval_loss": 2.2921693325042725,
|
| 1942 |
+
"eval_runtime": 246.0396,
|
| 1943 |
+
"eval_samples_per_second": 812.877,
|
| 1944 |
+
"eval_steps_per_second": 50.805,
|
| 1945 |
+
"step": 1408000
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 5.96,
|
| 1949 |
+
"eval_loss": 2.2846784591674805,
|
| 1950 |
+
"eval_runtime": 243.9013,
|
| 1951 |
+
"eval_samples_per_second": 820.004,
|
| 1952 |
+
"eval_steps_per_second": 51.25,
|
| 1953 |
+
"step": 1416000
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 6.0,
|
| 1957 |
+
"learning_rate": 1.6673333333333333e-07,
|
| 1958 |
+
"loss": 2.4291,
|
| 1959 |
+
"step": 1424000
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 6.0,
|
| 1963 |
+
"eval_loss": 2.285550117492676,
|
| 1964 |
+
"eval_runtime": 244.514,
|
| 1965 |
+
"eval_samples_per_second": 817.949,
|
| 1966 |
+
"eval_steps_per_second": 51.122,
|
| 1967 |
+
"step": 1424000
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 6.03,
|
| 1971 |
+
"eval_loss": 2.287602186203003,
|
| 1972 |
+
"eval_runtime": 245.7265,
|
| 1973 |
+
"eval_samples_per_second": 813.913,
|
| 1974 |
+
"eval_steps_per_second": 50.87,
|
| 1975 |
+
"step": 1432000
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 6.06,
|
| 1979 |
+
"learning_rate": 1.64e-07,
|
| 1980 |
+
"loss": 2.4289,
|
| 1981 |
+
"step": 1440000
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 6.06,
|
| 1985 |
+
"eval_loss": 2.282216787338257,
|
| 1986 |
+
"eval_runtime": 247.3389,
|
| 1987 |
+
"eval_samples_per_second": 808.607,
|
| 1988 |
+
"eval_steps_per_second": 50.538,
|
| 1989 |
+
"step": 1440000
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 6.1,
|
| 1993 |
+
"eval_loss": 2.2787294387817383,
|
| 1994 |
+
"eval_runtime": 247.7887,
|
| 1995 |
+
"eval_samples_per_second": 807.139,
|
| 1996 |
+
"eval_steps_per_second": 50.446,
|
| 1997 |
+
"step": 1448000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 6.13,
|
| 2001 |
+
"learning_rate": 1.6126666666666667e-07,
|
| 2002 |
+
"loss": 2.4272,
|
| 2003 |
+
"step": 1456000
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 6.13,
|
| 2007 |
+
"eval_loss": 2.2810943126678467,
|
| 2008 |
+
"eval_runtime": 244.5008,
|
| 2009 |
+
"eval_samples_per_second": 817.993,
|
| 2010 |
+
"eval_steps_per_second": 51.125,
|
| 2011 |
+
"step": 1456000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 6.16,
|
| 2015 |
+
"eval_loss": 2.285348415374756,
|
| 2016 |
+
"eval_runtime": 246.9617,
|
| 2017 |
+
"eval_samples_per_second": 809.842,
|
| 2018 |
+
"eval_steps_per_second": 50.615,
|
| 2019 |
+
"step": 1464000
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 6.2,
|
| 2023 |
+
"learning_rate": 1.5853333333333332e-07,
|
| 2024 |
+
"loss": 2.4267,
|
| 2025 |
+
"step": 1472000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 6.2,
|
| 2029 |
+
"eval_loss": 2.2818119525909424,
|
| 2030 |
+
"eval_runtime": 245.9708,
|
| 2031 |
+
"eval_samples_per_second": 813.105,
|
| 2032 |
+
"eval_steps_per_second": 50.819,
|
| 2033 |
+
"step": 1472000
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 6.23,
|
| 2037 |
+
"eval_loss": 2.2764692306518555,
|
| 2038 |
+
"eval_runtime": 249.5295,
|
| 2039 |
+
"eval_samples_per_second": 801.509,
|
| 2040 |
+
"eval_steps_per_second": 50.094,
|
| 2041 |
+
"step": 1480000
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 6.27,
|
| 2045 |
+
"learning_rate": 1.558e-07,
|
| 2046 |
+
"loss": 2.4237,
|
| 2047 |
+
"step": 1488000
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 6.27,
|
| 2051 |
+
"eval_loss": 2.2790732383728027,
|
| 2052 |
+
"eval_runtime": 247.115,
|
| 2053 |
+
"eval_samples_per_second": 809.34,
|
| 2054 |
+
"eval_steps_per_second": 50.584,
|
| 2055 |
+
"step": 1488000
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 6.3,
|
| 2059 |
+
"eval_loss": 2.2768290042877197,
|
| 2060 |
+
"eval_runtime": 246.6491,
|
| 2061 |
+
"eval_samples_per_second": 810.868,
|
| 2062 |
+
"eval_steps_per_second": 50.679,
|
| 2063 |
+
"step": 1496000
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 6.33,
|
| 2067 |
+
"learning_rate": 1.5306666666666666e-07,
|
| 2068 |
+
"loss": 2.4277,
|
| 2069 |
+
"step": 1504000
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 6.33,
|
| 2073 |
+
"eval_loss": 2.286600351333618,
|
| 2074 |
+
"eval_runtime": 247.5007,
|
| 2075 |
+
"eval_samples_per_second": 808.078,
|
| 2076 |
+
"eval_steps_per_second": 50.505,
|
| 2077 |
+
"step": 1504000
|
| 2078 |
+
},
|
| 2079 |
+
{
|
| 2080 |
+
"epoch": 6.37,
|
| 2081 |
+
"eval_loss": 2.282073974609375,
|
| 2082 |
+
"eval_runtime": 245.5734,
|
| 2083 |
+
"eval_samples_per_second": 814.42,
|
| 2084 |
+
"eval_steps_per_second": 50.901,
|
| 2085 |
+
"step": 1512000
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 6.4,
|
| 2089 |
+
"learning_rate": 1.5033333333333332e-07,
|
| 2090 |
+
"loss": 2.4316,
|
| 2091 |
+
"step": 1520000
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 6.4,
|
| 2095 |
+
"eval_loss": 2.285618305206299,
|
| 2096 |
+
"eval_runtime": 245.7388,
|
| 2097 |
+
"eval_samples_per_second": 813.872,
|
| 2098 |
+
"eval_steps_per_second": 50.867,
|
| 2099 |
+
"step": 1520000
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 6.43,
|
| 2103 |
+
"eval_loss": 2.2820215225219727,
|
| 2104 |
+
"eval_runtime": 247.4957,
|
| 2105 |
+
"eval_samples_per_second": 808.095,
|
| 2106 |
+
"eval_steps_per_second": 50.506,
|
| 2107 |
+
"step": 1528000
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 6.47,
|
| 2111 |
+
"learning_rate": 1.476e-07,
|
| 2112 |
+
"loss": 2.4222,
|
| 2113 |
+
"step": 1536000
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 6.47,
|
| 2117 |
+
"eval_loss": 2.2891499996185303,
|
| 2118 |
+
"eval_runtime": 245.6614,
|
| 2119 |
+
"eval_samples_per_second": 814.129,
|
| 2120 |
+
"eval_steps_per_second": 50.883,
|
| 2121 |
+
"step": 1536000
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 6.5,
|
| 2125 |
+
"eval_loss": 2.2802817821502686,
|
| 2126 |
+
"eval_runtime": 246.4523,
|
| 2127 |
+
"eval_samples_per_second": 811.516,
|
| 2128 |
+
"eval_steps_per_second": 50.72,
|
| 2129 |
+
"step": 1544000
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 6.53,
|
| 2133 |
+
"learning_rate": 1.4486666666666665e-07,
|
| 2134 |
+
"loss": 2.426,
|
| 2135 |
+
"step": 1552000
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 6.53,
|
| 2139 |
+
"eval_loss": 2.279686689376831,
|
| 2140 |
+
"eval_runtime": 247.3876,
|
| 2141 |
+
"eval_samples_per_second": 808.448,
|
| 2142 |
+
"eval_steps_per_second": 50.528,
|
| 2143 |
+
"step": 1552000
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 6.57,
|
| 2147 |
+
"eval_loss": 2.284360408782959,
|
| 2148 |
+
"eval_runtime": 246.8856,
|
| 2149 |
+
"eval_samples_per_second": 810.092,
|
| 2150 |
+
"eval_steps_per_second": 50.631,
|
| 2151 |
+
"step": 1560000
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 6.6,
|
| 2155 |
+
"learning_rate": 1.4213333333333334e-07,
|
| 2156 |
+
"loss": 2.422,
|
| 2157 |
+
"step": 1568000
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 6.6,
|
| 2161 |
+
"eval_loss": 2.287179708480835,
|
| 2162 |
+
"eval_runtime": 245.0018,
|
| 2163 |
+
"eval_samples_per_second": 816.321,
|
| 2164 |
+
"eval_steps_per_second": 51.02,
|
| 2165 |
+
"step": 1568000
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 6.64,
|
| 2169 |
+
"eval_loss": 2.2903947830200195,
|
| 2170 |
+
"eval_runtime": 244.8984,
|
| 2171 |
+
"eval_samples_per_second": 816.665,
|
| 2172 |
+
"eval_steps_per_second": 51.042,
|
| 2173 |
+
"step": 1576000
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 6.67,
|
| 2177 |
+
"learning_rate": 1.3940000000000002e-07,
|
| 2178 |
+
"loss": 2.4323,
|
| 2179 |
+
"step": 1584000
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 6.67,
|
| 2183 |
+
"eval_loss": 2.2796852588653564,
|
| 2184 |
+
"eval_runtime": 245.135,
|
| 2185 |
+
"eval_samples_per_second": 815.877,
|
| 2186 |
+
"eval_steps_per_second": 50.992,
|
| 2187 |
+
"step": 1584000
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 6.7,
|
| 2191 |
+
"eval_loss": 2.275681495666504,
|
| 2192 |
+
"eval_runtime": 247.5616,
|
| 2193 |
+
"eval_samples_per_second": 807.88,
|
| 2194 |
+
"eval_steps_per_second": 50.492,
|
| 2195 |
+
"step": 1592000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 6.74,
|
| 2199 |
+
"learning_rate": 1.3666666666666665e-07,
|
| 2200 |
+
"loss": 2.4315,
|
| 2201 |
+
"step": 1600000
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 6.74,
|
| 2205 |
+
"eval_loss": 2.287421703338623,
|
| 2206 |
+
"eval_runtime": 247.5879,
|
| 2207 |
+
"eval_samples_per_second": 807.794,
|
| 2208 |
+
"eval_steps_per_second": 50.487,
|
| 2209 |
+
"step": 1600000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 6.77,
|
| 2213 |
+
"eval_loss": 2.276327133178711,
|
| 2214 |
+
"eval_runtime": 245.1196,
|
| 2215 |
+
"eval_samples_per_second": 815.928,
|
| 2216 |
+
"eval_steps_per_second": 50.996,
|
| 2217 |
+
"step": 1608000
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 6.8,
|
| 2221 |
+
"learning_rate": 1.3393333333333333e-07,
|
| 2222 |
+
"loss": 2.421,
|
| 2223 |
+
"step": 1616000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 6.8,
|
| 2227 |
+
"eval_loss": 2.2857308387756348,
|
| 2228 |
+
"eval_runtime": 245.1941,
|
| 2229 |
+
"eval_samples_per_second": 815.68,
|
| 2230 |
+
"eval_steps_per_second": 50.98,
|
| 2231 |
+
"step": 1616000
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 6.84,
|
| 2235 |
+
"eval_loss": 2.280374765396118,
|
| 2236 |
+
"eval_runtime": 247.8166,
|
| 2237 |
+
"eval_samples_per_second": 807.048,
|
| 2238 |
+
"eval_steps_per_second": 50.441,
|
| 2239 |
+
"step": 1624000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 6.87,
|
| 2243 |
+
"learning_rate": 1.312e-07,
|
| 2244 |
+
"loss": 2.4299,
|
| 2245 |
+
"step": 1632000
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 6.87,
|
| 2249 |
+
"eval_loss": 2.282515287399292,
|
| 2250 |
+
"eval_runtime": 245.4773,
|
| 2251 |
+
"eval_samples_per_second": 814.739,
|
| 2252 |
+
"eval_steps_per_second": 50.921,
|
| 2253 |
+
"step": 1632000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 6.91,
|
| 2257 |
+
"eval_loss": 2.2819290161132812,
|
| 2258 |
+
"eval_runtime": 247.6539,
|
| 2259 |
+
"eval_samples_per_second": 807.579,
|
| 2260 |
+
"eval_steps_per_second": 50.474,
|
| 2261 |
+
"step": 1640000
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 6.94,
|
| 2265 |
+
"learning_rate": 1.2846666666666667e-07,
|
| 2266 |
+
"loss": 2.4289,
|
| 2267 |
+
"step": 1648000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 6.94,
|
| 2271 |
+
"eval_loss": 2.28240704536438,
|
| 2272 |
+
"eval_runtime": 245.8148,
|
| 2273 |
+
"eval_samples_per_second": 813.621,
|
| 2274 |
+
"eval_steps_per_second": 50.851,
|
| 2275 |
+
"step": 1648000
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 6.97,
|
| 2279 |
+
"eval_loss": 2.282073974609375,
|
| 2280 |
+
"eval_runtime": 247.9192,
|
| 2281 |
+
"eval_samples_per_second": 806.714,
|
| 2282 |
+
"eval_steps_per_second": 50.42,
|
| 2283 |
+
"step": 1656000
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 7.01,
|
| 2287 |
+
"learning_rate": 1.2573333333333332e-07,
|
| 2288 |
+
"loss": 2.4257,
|
| 2289 |
+
"step": 1664000
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 7.01,
|
| 2293 |
+
"eval_loss": 2.2802059650421143,
|
| 2294 |
+
"eval_runtime": 248.4226,
|
| 2295 |
+
"eval_samples_per_second": 805.08,
|
| 2296 |
+
"eval_steps_per_second": 50.317,
|
| 2297 |
+
"step": 1664000
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 7.04,
|
| 2301 |
+
"eval_loss": 2.2760419845581055,
|
| 2302 |
+
"eval_runtime": 246.2143,
|
| 2303 |
+
"eval_samples_per_second": 812.3,
|
| 2304 |
+
"eval_steps_per_second": 50.769,
|
| 2305 |
+
"step": 1672000
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 7.07,
|
| 2309 |
+
"learning_rate": 1.23e-07,
|
| 2310 |
+
"loss": 2.4227,
|
| 2311 |
+
"step": 1680000
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 7.07,
|
| 2315 |
+
"eval_loss": 2.28104305267334,
|
| 2316 |
+
"eval_runtime": 247.8478,
|
| 2317 |
+
"eval_samples_per_second": 806.947,
|
| 2318 |
+
"eval_steps_per_second": 50.434,
|
| 2319 |
+
"step": 1680000
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 7.11,
|
| 2323 |
+
"eval_loss": 2.2776732444763184,
|
| 2324 |
+
"eval_runtime": 245.7431,
|
| 2325 |
+
"eval_samples_per_second": 813.858,
|
| 2326 |
+
"eval_steps_per_second": 50.866,
|
| 2327 |
+
"step": 1688000
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 7.14,
|
| 2331 |
+
"learning_rate": 1.2026666666666666e-07,
|
| 2332 |
+
"loss": 2.4287,
|
| 2333 |
+
"step": 1696000
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 7.14,
|
| 2337 |
+
"eval_loss": 2.277235984802246,
|
| 2338 |
+
"eval_runtime": 245.5934,
|
| 2339 |
+
"eval_samples_per_second": 814.354,
|
| 2340 |
+
"eval_steps_per_second": 50.897,
|
| 2341 |
+
"step": 1696000
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 7.17,
|
| 2345 |
+
"eval_loss": 2.2786269187927246,
|
| 2346 |
+
"eval_runtime": 246.4934,
|
| 2347 |
+
"eval_samples_per_second": 811.381,
|
| 2348 |
+
"eval_steps_per_second": 50.711,
|
| 2349 |
+
"step": 1704000
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 7.21,
|
| 2353 |
+
"learning_rate": 1.1753333333333334e-07,
|
| 2354 |
+
"loss": 2.4227,
|
| 2355 |
+
"step": 1712000
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 7.21,
|
| 2359 |
+
"eval_loss": 2.285905361175537,
|
| 2360 |
+
"eval_runtime": 246.4981,
|
| 2361 |
+
"eval_samples_per_second": 811.365,
|
| 2362 |
+
"eval_steps_per_second": 50.71,
|
| 2363 |
+
"step": 1712000
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 7.24,
|
| 2367 |
+
"eval_loss": 2.2862019538879395,
|
| 2368 |
+
"eval_runtime": 247.148,
|
| 2369 |
+
"eval_samples_per_second": 809.232,
|
| 2370 |
+
"eval_steps_per_second": 50.577,
|
| 2371 |
+
"step": 1720000
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 7.28,
|
| 2375 |
+
"learning_rate": 1.1480000000000001e-07,
|
| 2376 |
+
"loss": 2.4262,
|
| 2377 |
+
"step": 1728000
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 7.28,
|
| 2381 |
+
"eval_loss": 2.278900146484375,
|
| 2382 |
+
"eval_runtime": 246.4039,
|
| 2383 |
+
"eval_samples_per_second": 811.675,
|
| 2384 |
+
"eval_steps_per_second": 50.73,
|
| 2385 |
+
"step": 1728000
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 7.31,
|
| 2389 |
+
"eval_loss": 2.2848124504089355,
|
| 2390 |
+
"eval_runtime": 246.3089,
|
| 2391 |
+
"eval_samples_per_second": 811.988,
|
| 2392 |
+
"eval_steps_per_second": 50.749,
|
| 2393 |
+
"step": 1736000
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 7.34,
|
| 2397 |
+
"learning_rate": 1.1206666666666666e-07,
|
| 2398 |
+
"loss": 2.4263,
|
| 2399 |
+
"step": 1744000
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 7.34,
|
| 2403 |
+
"eval_loss": 2.275432586669922,
|
| 2404 |
+
"eval_runtime": 246.8456,
|
| 2405 |
+
"eval_samples_per_second": 810.223,
|
| 2406 |
+
"eval_steps_per_second": 50.639,
|
| 2407 |
+
"step": 1744000
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 7.38,
|
| 2411 |
+
"eval_loss": 2.2777817249298096,
|
| 2412 |
+
"eval_runtime": 247.2848,
|
| 2413 |
+
"eval_samples_per_second": 808.784,
|
| 2414 |
+
"eval_steps_per_second": 50.549,
|
| 2415 |
+
"step": 1752000
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 7.41,
|
| 2419 |
+
"learning_rate": 1.0933333333333333e-07,
|
| 2420 |
+
"loss": 2.4246,
|
| 2421 |
+
"step": 1760000
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 7.41,
|
| 2425 |
+
"eval_loss": 2.27353572845459,
|
| 2426 |
+
"eval_runtime": 247.6065,
|
| 2427 |
+
"eval_samples_per_second": 807.733,
|
| 2428 |
+
"eval_steps_per_second": 50.483,
|
| 2429 |
+
"step": 1760000
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 7.44,
|
| 2433 |
+
"eval_loss": 2.2827255725860596,
|
| 2434 |
+
"eval_runtime": 246.5965,
|
| 2435 |
+
"eval_samples_per_second": 811.041,
|
| 2436 |
+
"eval_steps_per_second": 50.69,
|
| 2437 |
+
"step": 1768000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 7.48,
|
| 2441 |
+
"learning_rate": 1.066e-07,
|
| 2442 |
+
"loss": 2.4147,
|
| 2443 |
+
"step": 1776000
|
| 2444 |
+
},
|
| 2445 |
+
{
|
| 2446 |
+
"epoch": 7.48,
|
| 2447 |
+
"eval_loss": 2.285012722015381,
|
| 2448 |
+
"eval_runtime": 246.8181,
|
| 2449 |
+
"eval_samples_per_second": 810.313,
|
| 2450 |
+
"eval_steps_per_second": 50.645,
|
| 2451 |
+
"step": 1776000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 7.51,
|
| 2455 |
+
"eval_loss": 2.282083511352539,
|
| 2456 |
+
"eval_runtime": 246.9583,
|
| 2457 |
+
"eval_samples_per_second": 809.853,
|
| 2458 |
+
"eval_steps_per_second": 50.616,
|
| 2459 |
+
"step": 1784000
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 7.55,
|
| 2463 |
+
"learning_rate": 1.0386666666666667e-07,
|
| 2464 |
+
"loss": 2.4288,
|
| 2465 |
+
"step": 1792000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 7.55,
|
| 2469 |
+
"eval_loss": 2.280324935913086,
|
| 2470 |
+
"eval_runtime": 246.1706,
|
| 2471 |
+
"eval_samples_per_second": 812.445,
|
| 2472 |
+
"eval_steps_per_second": 50.778,
|
| 2473 |
+
"step": 1792000
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 7.58,
|
| 2477 |
+
"eval_loss": 2.2760393619537354,
|
| 2478 |
+
"eval_runtime": 246.0656,
|
| 2479 |
+
"eval_samples_per_second": 812.791,
|
| 2480 |
+
"eval_steps_per_second": 50.799,
|
| 2481 |
+
"step": 1800000
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 7.61,
|
| 2485 |
+
"learning_rate": 1.0113333333333334e-07,
|
| 2486 |
+
"loss": 2.4231,
|
| 2487 |
+
"step": 1808000
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 7.61,
|
| 2491 |
+
"eval_loss": 2.2748613357543945,
|
| 2492 |
+
"eval_runtime": 246.3683,
|
| 2493 |
+
"eval_samples_per_second": 811.793,
|
| 2494 |
+
"eval_steps_per_second": 50.737,
|
| 2495 |
+
"step": 1808000
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 7.65,
|
| 2499 |
+
"eval_loss": 2.2749080657958984,
|
| 2500 |
+
"eval_runtime": 246.3319,
|
| 2501 |
+
"eval_samples_per_second": 811.913,
|
| 2502 |
+
"eval_steps_per_second": 50.745,
|
| 2503 |
+
"step": 1816000
|
| 2504 |
+
},
|
| 2505 |
+
{
|
| 2506 |
+
"epoch": 7.68,
|
| 2507 |
+
"learning_rate": 9.84e-08,
|
| 2508 |
+
"loss": 2.4243,
|
| 2509 |
+
"step": 1824000
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 7.68,
|
| 2513 |
+
"eval_loss": 2.274254322052002,
|
| 2514 |
+
"eval_runtime": 245.7767,
|
| 2515 |
+
"eval_samples_per_second": 813.747,
|
| 2516 |
+
"eval_steps_per_second": 50.859,
|
| 2517 |
+
"step": 1824000
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 7.71,
|
| 2521 |
+
"eval_loss": 2.2792415618896484,
|
| 2522 |
+
"eval_runtime": 247.2515,
|
| 2523 |
+
"eval_samples_per_second": 808.893,
|
| 2524 |
+
"eval_steps_per_second": 50.556,
|
| 2525 |
+
"step": 1832000
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 7.75,
|
| 2529 |
+
"learning_rate": 9.566666666666666e-08,
|
| 2530 |
+
"loss": 2.4215,
|
| 2531 |
+
"step": 1840000
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 7.75,
|
| 2535 |
+
"eval_loss": 2.275153398513794,
|
| 2536 |
+
"eval_runtime": 247.4479,
|
| 2537 |
+
"eval_samples_per_second": 808.251,
|
| 2538 |
+
"eval_steps_per_second": 50.516,
|
| 2539 |
+
"step": 1840000
|
| 2540 |
+
},
|
| 2541 |
+
{
|
| 2542 |
+
"epoch": 7.78,
|
| 2543 |
+
"eval_loss": 2.2769758701324463,
|
| 2544 |
+
"eval_runtime": 247.8276,
|
| 2545 |
+
"eval_samples_per_second": 807.013,
|
| 2546 |
+
"eval_steps_per_second": 50.438,
|
| 2547 |
+
"step": 1848000
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 7.81,
|
| 2551 |
+
"learning_rate": 9.293333333333333e-08,
|
| 2552 |
+
"loss": 2.4213,
|
| 2553 |
+
"step": 1856000
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 7.81,
|
| 2557 |
+
"eval_loss": 2.280165672302246,
|
| 2558 |
+
"eval_runtime": 246.5191,
|
| 2559 |
+
"eval_samples_per_second": 811.296,
|
| 2560 |
+
"eval_steps_per_second": 50.706,
|
| 2561 |
+
"step": 1856000
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 7.85,
|
| 2565 |
+
"eval_loss": 2.27955961227417,
|
| 2566 |
+
"eval_runtime": 248.3255,
|
| 2567 |
+
"eval_samples_per_second": 805.394,
|
| 2568 |
+
"eval_steps_per_second": 50.337,
|
| 2569 |
+
"step": 1864000
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 7.88,
|
| 2573 |
+
"learning_rate": 9.02e-08,
|
| 2574 |
+
"loss": 2.4236,
|
| 2575 |
+
"step": 1872000
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 7.88,
|
| 2579 |
+
"eval_loss": 2.2882931232452393,
|
| 2580 |
+
"eval_runtime": 246.4462,
|
| 2581 |
+
"eval_samples_per_second": 811.536,
|
| 2582 |
+
"eval_steps_per_second": 50.721,
|
| 2583 |
+
"step": 1872000
|
| 2584 |
+
},
|
| 2585 |
+
{
|
| 2586 |
+
"epoch": 7.92,
|
| 2587 |
+
"eval_loss": 2.279242515563965,
|
| 2588 |
+
"eval_runtime": 246.5281,
|
| 2589 |
+
"eval_samples_per_second": 811.266,
|
| 2590 |
+
"eval_steps_per_second": 50.704,
|
| 2591 |
+
"step": 1880000
|
| 2592 |
+
},
|
| 2593 |
+
{
|
| 2594 |
+
"epoch": 7.95,
|
| 2595 |
+
"learning_rate": 8.746666666666667e-08,
|
| 2596 |
+
"loss": 2.4237,
|
| 2597 |
+
"step": 1888000
|
| 2598 |
+
},
|
| 2599 |
+
{
|
| 2600 |
+
"epoch": 7.95,
|
| 2601 |
+
"eval_loss": 2.272642135620117,
|
| 2602 |
+
"eval_runtime": 247.5094,
|
| 2603 |
+
"eval_samples_per_second": 808.05,
|
| 2604 |
+
"eval_steps_per_second": 50.503,
|
| 2605 |
+
"step": 1888000
|
| 2606 |
+
},
|
| 2607 |
+
{
|
| 2608 |
+
"epoch": 7.98,
|
| 2609 |
+
"eval_loss": 2.2816178798675537,
|
| 2610 |
+
"eval_runtime": 248.3802,
|
| 2611 |
+
"eval_samples_per_second": 805.217,
|
| 2612 |
+
"eval_steps_per_second": 50.326,
|
| 2613 |
+
"step": 1896000
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 8.02,
|
| 2617 |
+
"learning_rate": 8.473333333333334e-08,
|
| 2618 |
+
"loss": 2.4183,
|
| 2619 |
+
"step": 1904000
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 8.02,
|
| 2623 |
+
"eval_loss": 2.2790186405181885,
|
| 2624 |
+
"eval_runtime": 247.7385,
|
| 2625 |
+
"eval_samples_per_second": 807.303,
|
| 2626 |
+
"eval_steps_per_second": 50.456,
|
| 2627 |
+
"step": 1904000
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 8.05,
|
| 2631 |
+
"eval_loss": 2.2814745903015137,
|
| 2632 |
+
"eval_runtime": 248.3044,
|
| 2633 |
+
"eval_samples_per_second": 805.463,
|
| 2634 |
+
"eval_steps_per_second": 50.341,
|
| 2635 |
+
"step": 1912000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 8.08,
|
| 2639 |
+
"learning_rate": 8.2e-08,
|
| 2640 |
+
"loss": 2.4215,
|
| 2641 |
+
"step": 1920000
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 8.08,
|
| 2645 |
+
"eval_loss": 2.277374744415283,
|
| 2646 |
+
"eval_runtime": 246.8648,
|
| 2647 |
+
"eval_samples_per_second": 810.16,
|
| 2648 |
+
"eval_steps_per_second": 50.635,
|
| 2649 |
+
"step": 1920000
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 8.12,
|
| 2653 |
+
"eval_loss": 2.270047664642334,
|
| 2654 |
+
"eval_runtime": 246.8413,
|
| 2655 |
+
"eval_samples_per_second": 810.237,
|
| 2656 |
+
"eval_steps_per_second": 50.64,
|
| 2657 |
+
"step": 1928000
|
| 2658 |
+
},
|
| 2659 |
+
{
|
| 2660 |
+
"epoch": 8.15,
|
| 2661 |
+
"learning_rate": 7.926666666666666e-08,
|
| 2662 |
+
"loss": 2.4258,
|
| 2663 |
+
"step": 1936000
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 8.15,
|
| 2667 |
+
"eval_loss": 2.276344060897827,
|
| 2668 |
+
"eval_runtime": 246.9748,
|
| 2669 |
+
"eval_samples_per_second": 809.799,
|
| 2670 |
+
"eval_steps_per_second": 50.612,
|
| 2671 |
+
"step": 1936000
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 8.19,
|
| 2675 |
+
"eval_loss": 2.278593063354492,
|
| 2676 |
+
"eval_runtime": 248.7283,
|
| 2677 |
+
"eval_samples_per_second": 804.09,
|
| 2678 |
+
"eval_steps_per_second": 50.256,
|
| 2679 |
+
"step": 1944000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 8.22,
|
| 2683 |
+
"learning_rate": 7.653333333333333e-08,
|
| 2684 |
+
"loss": 2.4209,
|
| 2685 |
+
"step": 1952000
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 8.22,
|
| 2689 |
+
"eval_loss": 2.2763326168060303,
|
| 2690 |
+
"eval_runtime": 247.5692,
|
| 2691 |
+
"eval_samples_per_second": 807.855,
|
| 2692 |
+
"eval_steps_per_second": 50.491,
|
| 2693 |
+
"step": 1952000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 8.25,
|
| 2697 |
+
"eval_loss": 2.278881072998047,
|
| 2698 |
+
"eval_runtime": 247.8151,
|
| 2699 |
+
"eval_samples_per_second": 807.053,
|
| 2700 |
+
"eval_steps_per_second": 50.441,
|
| 2701 |
+
"step": 1960000
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 8.29,
|
| 2705 |
+
"learning_rate": 7.38e-08,
|
| 2706 |
+
"loss": 2.4217,
|
| 2707 |
+
"step": 1968000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 8.29,
|
| 2711 |
+
"eval_loss": 2.2783920764923096,
|
| 2712 |
+
"eval_runtime": 248.9593,
|
| 2713 |
+
"eval_samples_per_second": 803.344,
|
| 2714 |
+
"eval_steps_per_second": 50.209,
|
| 2715 |
+
"step": 1968000
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 8.32,
|
| 2719 |
+
"eval_loss": 2.2773079872131348,
|
| 2720 |
+
"eval_runtime": 247.6818,
|
| 2721 |
+
"eval_samples_per_second": 807.488,
|
| 2722 |
+
"eval_steps_per_second": 50.468,
|
| 2723 |
+
"step": 1976000
|
| 2724 |
+
},
|
| 2725 |
+
{
|
| 2726 |
+
"epoch": 8.35,
|
| 2727 |
+
"learning_rate": 7.106666666666667e-08,
|
| 2728 |
+
"loss": 2.4279,
|
| 2729 |
+
"step": 1984000
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 8.35,
|
| 2733 |
+
"eval_loss": 2.286058187484741,
|
| 2734 |
+
"eval_runtime": 247.013,
|
| 2735 |
+
"eval_samples_per_second": 809.674,
|
| 2736 |
+
"eval_steps_per_second": 50.605,
|
| 2737 |
+
"step": 1984000
|
| 2738 |
+
},
|
| 2739 |
+
{
|
| 2740 |
+
"epoch": 8.39,
|
| 2741 |
+
"eval_loss": 2.2728214263916016,
|
| 2742 |
+
"eval_runtime": 247.152,
|
| 2743 |
+
"eval_samples_per_second": 809.219,
|
| 2744 |
+
"eval_steps_per_second": 50.576,
|
| 2745 |
+
"step": 1992000
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 8.42,
|
| 2749 |
+
"learning_rate": 6.833333333333332e-08,
|
| 2750 |
+
"loss": 2.4268,
|
| 2751 |
+
"step": 2000000
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 8.42,
|
| 2755 |
+
"eval_loss": 2.276221513748169,
|
| 2756 |
+
"eval_runtime": 247.3718,
|
| 2757 |
+
"eval_samples_per_second": 808.5,
|
| 2758 |
+
"eval_steps_per_second": 50.531,
|
| 2759 |
+
"step": 2000000
|
| 2760 |
+
},
|
| 2761 |
+
{
|
| 2762 |
+
"epoch": 8.45,
|
| 2763 |
+
"eval_loss": 2.2788944244384766,
|
| 2764 |
+
"eval_runtime": 247.8322,
|
| 2765 |
+
"eval_samples_per_second": 806.998,
|
| 2766 |
+
"eval_steps_per_second": 50.437,
|
| 2767 |
+
"step": 2008000
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 8.49,
|
| 2771 |
+
"learning_rate": 6.56e-08,
|
| 2772 |
+
"loss": 2.4177,
|
| 2773 |
+
"step": 2016000
|
| 2774 |
+
},
|
| 2775 |
+
{
|
| 2776 |
+
"epoch": 8.49,
|
| 2777 |
+
"eval_loss": 2.282210350036621,
|
| 2778 |
+
"eval_runtime": 247.9565,
|
| 2779 |
+
"eval_samples_per_second": 806.593,
|
| 2780 |
+
"eval_steps_per_second": 50.412,
|
| 2781 |
+
"step": 2016000
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 8.52,
|
| 2785 |
+
"eval_loss": 2.275873899459839,
|
| 2786 |
+
"eval_runtime": 247.5408,
|
| 2787 |
+
"eval_samples_per_second": 807.948,
|
| 2788 |
+
"eval_steps_per_second": 50.497,
|
| 2789 |
+
"step": 2024000
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 8.56,
|
| 2793 |
+
"learning_rate": 6.286666666666666e-08,
|
| 2794 |
+
"loss": 2.4166,
|
| 2795 |
+
"step": 2032000
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 8.56,
|
| 2799 |
+
"eval_loss": 2.2791593074798584,
|
| 2800 |
+
"eval_runtime": 248.9215,
|
| 2801 |
+
"eval_samples_per_second": 803.466,
|
| 2802 |
+
"eval_steps_per_second": 50.217,
|
| 2803 |
+
"step": 2032000
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 8.59,
|
| 2807 |
+
"eval_loss": 2.272076368331909,
|
| 2808 |
+
"eval_runtime": 248.0422,
|
| 2809 |
+
"eval_samples_per_second": 806.315,
|
| 2810 |
+
"eval_steps_per_second": 50.395,
|
| 2811 |
+
"step": 2040000
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 8.62,
|
| 2815 |
+
"learning_rate": 6.013333333333333e-08,
|
| 2816 |
+
"loss": 2.4223,
|
| 2817 |
+
"step": 2048000
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 8.62,
|
| 2821 |
+
"eval_loss": 2.2768054008483887,
|
| 2822 |
+
"eval_runtime": 247.9612,
|
| 2823 |
+
"eval_samples_per_second": 806.578,
|
| 2824 |
+
"eval_steps_per_second": 50.411,
|
| 2825 |
+
"step": 2048000
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 8.66,
|
| 2829 |
+
"eval_loss": 2.2726008892059326,
|
| 2830 |
+
"eval_runtime": 247.7176,
|
| 2831 |
+
"eval_samples_per_second": 807.371,
|
| 2832 |
+
"eval_steps_per_second": 50.461,
|
| 2833 |
+
"step": 2056000
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 8.69,
|
| 2837 |
+
"learning_rate": 5.7400000000000004e-08,
|
| 2838 |
+
"loss": 2.4139,
|
| 2839 |
+
"step": 2064000
|
| 2840 |
+
},
|
| 2841 |
+
{
|
| 2842 |
+
"epoch": 8.69,
|
| 2843 |
+
"eval_loss": 2.282505512237549,
|
| 2844 |
+
"eval_runtime": 247.6026,
|
| 2845 |
+
"eval_samples_per_second": 807.746,
|
| 2846 |
+
"eval_steps_per_second": 50.484,
|
| 2847 |
+
"step": 2064000
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 8.72,
|
| 2851 |
+
"eval_loss": 2.2738897800445557,
|
| 2852 |
+
"eval_runtime": 248.4813,
|
| 2853 |
+
"eval_samples_per_second": 804.89,
|
| 2854 |
+
"eval_steps_per_second": 50.306,
|
| 2855 |
+
"step": 2072000
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 8.76,
|
| 2859 |
+
"learning_rate": 5.4666666666666666e-08,
|
| 2860 |
+
"loss": 2.4236,
|
| 2861 |
+
"step": 2080000
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 8.76,
|
| 2865 |
+
"eval_loss": 2.2834410667419434,
|
| 2866 |
+
"eval_runtime": 248.3802,
|
| 2867 |
+
"eval_samples_per_second": 805.217,
|
| 2868 |
+
"eval_steps_per_second": 50.326,
|
| 2869 |
+
"step": 2080000
|
| 2870 |
+
},
|
| 2871 |
+
{
|
| 2872 |
+
"epoch": 8.79,
|
| 2873 |
+
"eval_loss": 2.2750093936920166,
|
| 2874 |
+
"eval_runtime": 247.5076,
|
| 2875 |
+
"eval_samples_per_second": 808.056,
|
| 2876 |
+
"eval_steps_per_second": 50.503,
|
| 2877 |
+
"step": 2088000
|
| 2878 |
+
},
|
| 2879 |
+
{
|
| 2880 |
+
"epoch": 8.83,
|
| 2881 |
+
"learning_rate": 5.1933333333333335e-08,
|
| 2882 |
+
"loss": 2.4235,
|
| 2883 |
+
"step": 2096000
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 8.83,
|
| 2887 |
+
"eval_loss": 2.275216817855835,
|
| 2888 |
+
"eval_runtime": 248.1106,
|
| 2889 |
+
"eval_samples_per_second": 806.092,
|
| 2890 |
+
"eval_steps_per_second": 50.381,
|
| 2891 |
+
"step": 2096000
|
| 2892 |
+
},
|
| 2893 |
+
{
|
| 2894 |
+
"epoch": 8.86,
|
| 2895 |
+
"eval_loss": 2.28031587600708,
|
| 2896 |
+
"eval_runtime": 247.6614,
|
| 2897 |
+
"eval_samples_per_second": 807.554,
|
| 2898 |
+
"eval_steps_per_second": 50.472,
|
| 2899 |
+
"step": 2104000
|
| 2900 |
+
},
|
| 2901 |
+
{
|
| 2902 |
+
"epoch": 8.89,
|
| 2903 |
+
"learning_rate": 4.92e-08,
|
| 2904 |
+
"loss": 2.4193,
|
| 2905 |
+
"step": 2112000
|
| 2906 |
+
},
|
| 2907 |
+
{
|
| 2908 |
+
"epoch": 8.89,
|
| 2909 |
+
"eval_loss": 2.2762908935546875,
|
| 2910 |
+
"eval_runtime": 247.8319,
|
| 2911 |
+
"eval_samples_per_second": 806.999,
|
| 2912 |
+
"eval_steps_per_second": 50.437,
|
| 2913 |
+
"step": 2112000
|
| 2914 |
+
},
|
| 2915 |
+
{
|
| 2916 |
+
"epoch": 8.93,
|
| 2917 |
+
"eval_loss": 2.2754592895507812,
|
| 2918 |
+
"eval_runtime": 247.5134,
|
| 2919 |
+
"eval_samples_per_second": 808.037,
|
| 2920 |
+
"eval_steps_per_second": 50.502,
|
| 2921 |
+
"step": 2120000
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 8.96,
|
| 2925 |
+
"learning_rate": 4.6466666666666666e-08,
|
| 2926 |
+
"loss": 2.4179,
|
| 2927 |
+
"step": 2128000
|
| 2928 |
+
},
|
| 2929 |
+
{
|
| 2930 |
+
"epoch": 8.96,
|
| 2931 |
+
"eval_loss": 2.279421091079712,
|
| 2932 |
+
"eval_runtime": 248.0477,
|
| 2933 |
+
"eval_samples_per_second": 806.297,
|
| 2934 |
+
"eval_steps_per_second": 50.394,
|
| 2935 |
+
"step": 2128000
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 8.99,
|
| 2939 |
+
"eval_loss": 2.2710611820220947,
|
| 2940 |
+
"eval_runtime": 247.3731,
|
| 2941 |
+
"eval_samples_per_second": 808.495,
|
| 2942 |
+
"eval_steps_per_second": 50.531,
|
| 2943 |
+
"step": 2136000
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 9.03,
|
| 2947 |
+
"learning_rate": 4.3733333333333335e-08,
|
| 2948 |
+
"loss": 2.4181,
|
| 2949 |
+
"step": 2144000
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 9.03,
|
| 2953 |
+
"eval_loss": 2.279233932495117,
|
| 2954 |
+
"eval_runtime": 248.7527,
|
| 2955 |
+
"eval_samples_per_second": 804.011,
|
| 2956 |
+
"eval_steps_per_second": 50.251,
|
| 2957 |
+
"step": 2144000
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 9.06,
|
| 2961 |
+
"eval_loss": 2.275193691253662,
|
| 2962 |
+
"eval_runtime": 248.241,
|
| 2963 |
+
"eval_samples_per_second": 805.669,
|
| 2964 |
+
"eval_steps_per_second": 50.354,
|
| 2965 |
+
"step": 2152000
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 9.09,
|
| 2969 |
+
"learning_rate": 4.1e-08,
|
| 2970 |
+
"loss": 2.4173,
|
| 2971 |
+
"step": 2160000
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 9.09,
|
| 2975 |
+
"eval_loss": 2.2775352001190186,
|
| 2976 |
+
"eval_runtime": 248.2377,
|
| 2977 |
+
"eval_samples_per_second": 805.679,
|
| 2978 |
+
"eval_steps_per_second": 50.355,
|
| 2979 |
+
"step": 2160000
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 9.13,
|
| 2983 |
+
"eval_loss": 2.2751924991607666,
|
| 2984 |
+
"eval_runtime": 248.4267,
|
| 2985 |
+
"eval_samples_per_second": 805.066,
|
| 2986 |
+
"eval_steps_per_second": 50.317,
|
| 2987 |
+
"step": 2168000
|
| 2988 |
+
},
|
| 2989 |
+
{
|
| 2990 |
+
"epoch": 9.16,
|
| 2991 |
+
"learning_rate": 3.8266666666666665e-08,
|
| 2992 |
+
"loss": 2.4242,
|
| 2993 |
+
"step": 2176000
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 9.16,
|
| 2997 |
+
"eval_loss": 2.272923469543457,
|
| 2998 |
+
"eval_runtime": 248.2584,
|
| 2999 |
+
"eval_samples_per_second": 805.612,
|
| 3000 |
+
"eval_steps_per_second": 50.351,
|
| 3001 |
+
"step": 2176000
|
| 3002 |
+
},
|
| 3003 |
+
{
|
| 3004 |
+
"epoch": 9.2,
|
| 3005 |
+
"eval_loss": 2.2793116569519043,
|
| 3006 |
+
"eval_runtime": 248.8342,
|
| 3007 |
+
"eval_samples_per_second": 803.748,
|
| 3008 |
+
"eval_steps_per_second": 50.234,
|
| 3009 |
+
"step": 2184000
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 9.23,
|
| 3013 |
+
"learning_rate": 3.5533333333333334e-08,
|
| 3014 |
+
"loss": 2.4166,
|
| 3015 |
+
"step": 2192000
|
| 3016 |
+
},
|
| 3017 |
+
{
|
| 3018 |
+
"epoch": 9.23,
|
| 3019 |
+
"eval_loss": 2.2719197273254395,
|
| 3020 |
+
"eval_runtime": 249.6006,
|
| 3021 |
+
"eval_samples_per_second": 801.28,
|
| 3022 |
+
"eval_steps_per_second": 50.08,
|
| 3023 |
+
"step": 2192000
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 9.26,
|
| 3027 |
+
"eval_loss": 2.281975507736206,
|
| 3028 |
+
"eval_runtime": 248.1074,
|
| 3029 |
+
"eval_samples_per_second": 806.103,
|
| 3030 |
+
"eval_steps_per_second": 50.381,
|
| 3031 |
+
"step": 2200000
|
| 3032 |
+
},
|
| 3033 |
+
{
|
| 3034 |
+
"epoch": 9.3,
|
| 3035 |
+
"learning_rate": 3.28e-08,
|
| 3036 |
+
"loss": 2.4181,
|
| 3037 |
+
"step": 2208000
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 9.3,
|
| 3041 |
+
"eval_loss": 2.2715933322906494,
|
| 3042 |
+
"eval_runtime": 249.4343,
|
| 3043 |
+
"eval_samples_per_second": 801.814,
|
| 3044 |
+
"eval_steps_per_second": 50.113,
|
| 3045 |
+
"step": 2208000
|
| 3046 |
+
},
|
| 3047 |
+
{
|
| 3048 |
+
"epoch": 9.33,
|
| 3049 |
+
"eval_loss": 2.285536050796509,
|
| 3050 |
+
"eval_runtime": 250.2462,
|
| 3051 |
+
"eval_samples_per_second": 799.213,
|
| 3052 |
+
"eval_steps_per_second": 49.951,
|
| 3053 |
+
"step": 2216000
|
| 3054 |
+
},
|
| 3055 |
+
{
|
| 3056 |
+
"epoch": 9.36,
|
| 3057 |
+
"learning_rate": 3.0066666666666665e-08,
|
| 3058 |
+
"loss": 2.4245,
|
| 3059 |
+
"step": 2224000
|
| 3060 |
+
},
|
| 3061 |
+
{
|
| 3062 |
+
"epoch": 9.36,
|
| 3063 |
+
"eval_loss": 2.2805118560791016,
|
| 3064 |
+
"eval_runtime": 248.8621,
|
| 3065 |
+
"eval_samples_per_second": 803.658,
|
| 3066 |
+
"eval_steps_per_second": 50.229,
|
| 3067 |
+
"step": 2224000
|
| 3068 |
+
},
|
| 3069 |
+
{
|
| 3070 |
+
"epoch": 9.4,
|
| 3071 |
+
"eval_loss": 2.272109270095825,
|
| 3072 |
+
"eval_runtime": 250.3664,
|
| 3073 |
+
"eval_samples_per_second": 798.829,
|
| 3074 |
+
"eval_steps_per_second": 49.927,
|
| 3075 |
+
"step": 2232000
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 9.43,
|
| 3079 |
+
"learning_rate": 2.7333333333333333e-08,
|
| 3080 |
+
"loss": 2.4204,
|
| 3081 |
+
"step": 2240000
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 9.43,
|
| 3085 |
+
"eval_loss": 2.270662546157837,
|
| 3086 |
+
"eval_runtime": 248.4336,
|
| 3087 |
+
"eval_samples_per_second": 805.044,
|
| 3088 |
+
"eval_steps_per_second": 50.315,
|
| 3089 |
+
"step": 2240000
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 9.47,
|
| 3093 |
+
"eval_loss": 2.2766847610473633,
|
| 3094 |
+
"eval_runtime": 248.6132,
|
| 3095 |
+
"eval_samples_per_second": 804.462,
|
| 3096 |
+
"eval_steps_per_second": 50.279,
|
| 3097 |
+
"step": 2248000
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 9.5,
|
| 3101 |
+
"learning_rate": 2.46e-08,
|
| 3102 |
+
"loss": 2.4255,
|
| 3103 |
+
"step": 2256000
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 9.5,
|
| 3107 |
+
"eval_loss": 2.2710325717926025,
|
| 3108 |
+
"eval_runtime": 248.6879,
|
| 3109 |
+
"eval_samples_per_second": 804.221,
|
| 3110 |
+
"eval_steps_per_second": 50.264,
|
| 3111 |
+
"step": 2256000
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 9.53,
|
| 3115 |
+
"eval_loss": 2.2814137935638428,
|
| 3116 |
+
"eval_runtime": 249.0563,
|
| 3117 |
+
"eval_samples_per_second": 803.031,
|
| 3118 |
+
"eval_steps_per_second": 50.189,
|
| 3119 |
+
"step": 2264000
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 9.57,
|
| 3123 |
+
"learning_rate": 2.1866666666666667e-08,
|
| 3124 |
+
"loss": 2.4254,
|
| 3125 |
+
"step": 2272000
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 9.57,
|
| 3129 |
+
"eval_loss": 2.274559259414673,
|
| 3130 |
+
"eval_runtime": 249.1887,
|
| 3131 |
+
"eval_samples_per_second": 802.605,
|
| 3132 |
+
"eval_steps_per_second": 50.163,
|
| 3133 |
+
"step": 2272000
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 9.6,
|
| 3137 |
+
"eval_loss": 2.2766289710998535,
|
| 3138 |
+
"eval_runtime": 248.6168,
|
| 3139 |
+
"eval_samples_per_second": 804.451,
|
| 3140 |
+
"eval_steps_per_second": 50.278,
|
| 3141 |
+
"step": 2280000
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 9.63,
|
| 3145 |
+
"learning_rate": 1.9133333333333333e-08,
|
| 3146 |
+
"loss": 2.4232,
|
| 3147 |
+
"step": 2288000
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 9.63,
|
| 3151 |
+
"eval_loss": 2.2724950313568115,
|
| 3152 |
+
"eval_runtime": 248.9678,
|
| 3153 |
+
"eval_samples_per_second": 803.317,
|
| 3154 |
+
"eval_steps_per_second": 50.207,
|
| 3155 |
+
"step": 2288000
|
| 3156 |
+
},
|
| 3157 |
+
{
|
| 3158 |
+
"epoch": 9.67,
|
| 3159 |
+
"eval_loss": 2.2764828205108643,
|
| 3160 |
+
"eval_runtime": 248.7636,
|
| 3161 |
+
"eval_samples_per_second": 803.976,
|
| 3162 |
+
"eval_steps_per_second": 50.249,
|
| 3163 |
+
"step": 2296000
|
| 3164 |
+
},
|
| 3165 |
+
{
|
| 3166 |
+
"epoch": 9.7,
|
| 3167 |
+
"learning_rate": 1.64e-08,
|
| 3168 |
+
"loss": 2.4189,
|
| 3169 |
+
"step": 2304000
|
| 3170 |
+
},
|
| 3171 |
+
{
|
| 3172 |
+
"epoch": 9.7,
|
| 3173 |
+
"eval_loss": 2.275614023208618,
|
| 3174 |
+
"eval_runtime": 249.9482,
|
| 3175 |
+
"eval_samples_per_second": 800.166,
|
| 3176 |
+
"eval_steps_per_second": 50.01,
|
| 3177 |
+
"step": 2304000
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 9.73,
|
| 3181 |
+
"eval_loss": 2.2767763137817383,
|
| 3182 |
+
"eval_runtime": 248.8725,
|
| 3183 |
+
"eval_samples_per_second": 803.624,
|
| 3184 |
+
"eval_steps_per_second": 50.227,
|
| 3185 |
+
"step": 2312000
|
| 3186 |
+
},
|
| 3187 |
+
{
|
| 3188 |
+
"epoch": 9.77,
|
| 3189 |
+
"learning_rate": 1.3666666666666667e-08,
|
| 3190 |
+
"loss": 2.4105,
|
| 3191 |
+
"step": 2320000
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 9.77,
|
| 3195 |
+
"eval_loss": 2.280364751815796,
|
| 3196 |
+
"eval_runtime": 248.8287,
|
| 3197 |
+
"eval_samples_per_second": 803.766,
|
| 3198 |
+
"eval_steps_per_second": 50.235,
|
| 3199 |
+
"step": 2320000
|
| 3200 |
+
},
|
| 3201 |
+
{
|
| 3202 |
+
"epoch": 9.8,
|
| 3203 |
+
"eval_loss": 2.2873153686523438,
|
| 3204 |
+
"eval_runtime": 248.9747,
|
| 3205 |
+
"eval_samples_per_second": 803.295,
|
| 3206 |
+
"eval_steps_per_second": 50.206,
|
| 3207 |
+
"step": 2328000
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 9.84,
|
| 3211 |
+
"learning_rate": 1.0933333333333334e-08,
|
| 3212 |
+
"loss": 2.415,
|
| 3213 |
+
"step": 2336000
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 9.84,
|
| 3217 |
+
"eval_loss": 2.278338670730591,
|
| 3218 |
+
"eval_runtime": 248.9743,
|
| 3219 |
+
"eval_samples_per_second": 803.296,
|
| 3220 |
+
"eval_steps_per_second": 50.206,
|
| 3221 |
+
"step": 2336000
|
| 3222 |
+
},
|
| 3223 |
+
{
|
| 3224 |
+
"epoch": 9.87,
|
| 3225 |
+
"eval_loss": 2.2737488746643066,
|
| 3226 |
+
"eval_runtime": 250.6409,
|
| 3227 |
+
"eval_samples_per_second": 797.954,
|
| 3228 |
+
"eval_steps_per_second": 49.872,
|
| 3229 |
+
"step": 2344000
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 9.9,
|
| 3233 |
+
"learning_rate": 8.2e-09,
|
| 3234 |
+
"loss": 2.4174,
|
| 3235 |
+
"step": 2352000
|
| 3236 |
+
},
|
| 3237 |
+
{
|
| 3238 |
+
"epoch": 9.9,
|
| 3239 |
+
"eval_loss": 2.278608798980713,
|
| 3240 |
+
"eval_runtime": 251.8413,
|
| 3241 |
+
"eval_samples_per_second": 794.151,
|
| 3242 |
+
"eval_steps_per_second": 49.634,
|
| 3243 |
+
"step": 2352000
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 9.94,
|
| 3247 |
+
"eval_loss": 2.2729651927948,
|
| 3248 |
+
"eval_runtime": 250.8498,
|
| 3249 |
+
"eval_samples_per_second": 797.29,
|
| 3250 |
+
"eval_steps_per_second": 49.831,
|
| 3251 |
+
"step": 2360000
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 9.97,
|
| 3255 |
+
"learning_rate": 5.466666666666667e-09,
|
| 3256 |
+
"loss": 2.4199,
|
| 3257 |
+
"step": 2368000
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 9.97,
|
| 3261 |
+
"eval_loss": 2.2793538570404053,
|
| 3262 |
+
"eval_runtime": 253.8866,
|
| 3263 |
+
"eval_samples_per_second": 787.753,
|
| 3264 |
+
"eval_steps_per_second": 49.235,
|
| 3265 |
+
"step": 2368000
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 10.0,
|
| 3269 |
+
"eval_loss": 2.284799337387085,
|
| 3270 |
+
"eval_runtime": 252.1889,
|
| 3271 |
+
"eval_samples_per_second": 793.056,
|
| 3272 |
+
"eval_steps_per_second": 49.566,
|
| 3273 |
+
"step": 2376000
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 10.04,
|
| 3277 |
+
"learning_rate": 2.7333333333333334e-09,
|
| 3278 |
+
"loss": 2.4224,
|
| 3279 |
+
"step": 2384000
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 10.04,
|
| 3283 |
+
"eval_loss": 2.2810616493225098,
|
| 3284 |
+
"eval_runtime": 251.6013,
|
| 3285 |
+
"eval_samples_per_second": 794.909,
|
| 3286 |
+
"eval_steps_per_second": 49.682,
|
| 3287 |
+
"step": 2384000
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 10.07,
|
| 3291 |
+
"eval_loss": 2.281796932220459,
|
| 3292 |
+
"eval_runtime": 251.0825,
|
| 3293 |
+
"eval_samples_per_second": 796.551,
|
| 3294 |
+
"eval_steps_per_second": 49.784,
|
| 3295 |
+
"step": 2392000
|
| 3296 |
+
},
|
| 3297 |
+
{
|
| 3298 |
+
"epoch": 10.11,
|
| 3299 |
+
"learning_rate": 0.0,
|
| 3300 |
+
"loss": 2.4226,
|
| 3301 |
+
"step": 2400000
|
| 3302 |
+
},
|
| 3303 |
+
{
|
| 3304 |
+
"epoch": 10.11,
|
| 3305 |
+
"eval_loss": 2.279843330383301,
|
| 3306 |
+
"eval_runtime": 250.0989,
|
| 3307 |
+
"eval_samples_per_second": 799.684,
|
| 3308 |
+
"eval_steps_per_second": 49.98,
|
| 3309 |
+
"step": 2400000
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 10.11,
|
| 3313 |
+
"step": 2400000,
|
| 3314 |
+
"total_flos": 8.368611666112401e+17,
|
| 3315 |
+
"train_loss": 2.4345372688802085,
|
| 3316 |
+
"train_runtime": 232385.3236,
|
| 3317 |
+
"train_samples_per_second": 165.243,
|
| 3318 |
+
"train_steps_per_second": 10.328
|
| 3319 |
+
}
|
| 3320 |
+
],
|
| 3321 |
+
"logging_steps": 16000,
|
| 3322 |
+
"max_steps": 2400000,
|
| 3323 |
+
"num_train_epochs": 11,
|
| 3324 |
+
"save_steps": 32000,
|
| 3325 |
+
"total_flos": 8.368611666112401e+17,
|
| 3326 |
+
"trial_name": null,
|
| 3327 |
+
"trial_params": null
|
| 3328 |
+
}
|