MHGanainy/mgpt-lora-multi-switzerland-imbalanced-1024
Browse files- README.md +2 -0
- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +3395 -0
README.md
CHANGED
|
@@ -15,6 +15,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
# mgpt-lora-multi-switzerland-imbalanced-1024
|
| 16 |
|
| 17 |
This model is a fine-tuned version of [ai-forever/mGPT](https://huggingface.co/ai-forever/mGPT) on an unknown dataset.
|
|
|
|
|
|
|
| 18 |
|
| 19 |
## Model description
|
| 20 |
|
|
|
|
| 15 |
# mgpt-lora-multi-switzerland-imbalanced-1024
|
| 16 |
|
| 17 |
This model is a fine-tuned version of [ai-forever/mGPT](https://huggingface.co/ai-forever/mGPT) on an unknown dataset.
|
| 18 |
+
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 1.7731
|
| 20 |
|
| 21 |
## Model description
|
| 22 |
|
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_loss": 1.7731051445007324,
|
| 4 |
+
"eval_runtime": 1215.8742,
|
| 5 |
+
"eval_samples_per_second": 78.599,
|
| 6 |
+
"eval_steps_per_second": 9.825,
|
| 7 |
+
"perplexity": 5.889111541886369,
|
| 8 |
+
"total_flos": 5.824755462670647e+18,
|
| 9 |
+
"train_loss": 2.213516155517568,
|
| 10 |
+
"train_runtime": 14249.5377,
|
| 11 |
+
"train_samples_per_second": 26.934,
|
| 12 |
+
"train_steps_per_second": 3.367
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_loss": 1.7731051445007324,
|
| 4 |
+
"eval_runtime": 1215.8742,
|
| 5 |
+
"eval_samples_per_second": 78.599,
|
| 6 |
+
"eval_steps_per_second": 9.825,
|
| 7 |
+
"perplexity": 5.889111541886369
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 5.824755462670647e+18,
|
| 4 |
+
"train_loss": 2.213516155517568,
|
| 5 |
+
"train_runtime": 14249.5377,
|
| 6 |
+
"train_samples_per_second": 26.934,
|
| 7 |
+
"train_steps_per_second": 3.367
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3395 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 47974,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.002084462417142619,
|
| 13 |
+
"grad_norm": 17.418737411499023,
|
| 14 |
+
"learning_rate": 5.2111831991453666e-08,
|
| 15 |
+
"loss": 6.7115,
|
| 16 |
+
"step": 100
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.004168924834285238,
|
| 20 |
+
"grad_norm": 14.615851402282715,
|
| 21 |
+
"learning_rate": 1.0422366398290733e-07,
|
| 22 |
+
"loss": 6.6983,
|
| 23 |
+
"step": 200
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.006253387251427857,
|
| 27 |
+
"grad_norm": 12.82895278930664,
|
| 28 |
+
"learning_rate": 1.5633549597436098e-07,
|
| 29 |
+
"loss": 6.6196,
|
| 30 |
+
"step": 300
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.008337849668570476,
|
| 34 |
+
"grad_norm": 9.737526893615723,
|
| 35 |
+
"learning_rate": 2.0844732796581466e-07,
|
| 36 |
+
"loss": 6.4232,
|
| 37 |
+
"step": 400
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.010422312085713094,
|
| 41 |
+
"grad_norm": 10.140837669372559,
|
| 42 |
+
"learning_rate": 2.605591599572683e-07,
|
| 43 |
+
"loss": 6.0683,
|
| 44 |
+
"step": 500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.012506774502855714,
|
| 48 |
+
"grad_norm": 6.8095598220825195,
|
| 49 |
+
"learning_rate": 3.1267099194872195e-07,
|
| 50 |
+
"loss": 5.6784,
|
| 51 |
+
"step": 600
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.014591236919998333,
|
| 55 |
+
"grad_norm": 4.862451076507568,
|
| 56 |
+
"learning_rate": 3.6478282394017564e-07,
|
| 57 |
+
"loss": 5.3257,
|
| 58 |
+
"step": 700
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.016675699337140953,
|
| 62 |
+
"grad_norm": 4.783125877380371,
|
| 63 |
+
"learning_rate": 4.168946559316293e-07,
|
| 64 |
+
"loss": 5.0905,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.01876016175428357,
|
| 69 |
+
"grad_norm": 2.310032367706299,
|
| 70 |
+
"learning_rate": 4.6900648792308296e-07,
|
| 71 |
+
"loss": 4.8735,
|
| 72 |
+
"step": 900
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.02084462417142619,
|
| 76 |
+
"grad_norm": 1.856097936630249,
|
| 77 |
+
"learning_rate": 5.211183199145366e-07,
|
| 78 |
+
"loss": 4.7671,
|
| 79 |
+
"step": 1000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.02292908658856881,
|
| 83 |
+
"grad_norm": 1.9036556482315063,
|
| 84 |
+
"learning_rate": 5.732301519059903e-07,
|
| 85 |
+
"loss": 4.6835,
|
| 86 |
+
"step": 1100
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.025013549005711427,
|
| 90 |
+
"grad_norm": 2.0767664909362793,
|
| 91 |
+
"learning_rate": 6.253419838974439e-07,
|
| 92 |
+
"loss": 4.5978,
|
| 93 |
+
"step": 1200
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.027098011422854045,
|
| 97 |
+
"grad_norm": 5.158967018127441,
|
| 98 |
+
"learning_rate": 6.774538158888977e-07,
|
| 99 |
+
"loss": 4.5468,
|
| 100 |
+
"step": 1300
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.029182473839996666,
|
| 104 |
+
"grad_norm": 2.4854819774627686,
|
| 105 |
+
"learning_rate": 7.295656478803513e-07,
|
| 106 |
+
"loss": 4.4797,
|
| 107 |
+
"step": 1400
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.031266936257139284,
|
| 111 |
+
"grad_norm": 2.1669373512268066,
|
| 112 |
+
"learning_rate": 7.81677479871805e-07,
|
| 113 |
+
"loss": 4.4619,
|
| 114 |
+
"step": 1500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.033351398674281905,
|
| 118 |
+
"grad_norm": 3.4315173625946045,
|
| 119 |
+
"learning_rate": 8.337893118632586e-07,
|
| 120 |
+
"loss": 4.4004,
|
| 121 |
+
"step": 1600
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.03543586109142452,
|
| 125 |
+
"grad_norm": 4.037099361419678,
|
| 126 |
+
"learning_rate": 8.859011438547123e-07,
|
| 127 |
+
"loss": 4.3784,
|
| 128 |
+
"step": 1700
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.03752032350856714,
|
| 132 |
+
"grad_norm": 1.5988932847976685,
|
| 133 |
+
"learning_rate": 9.380129758461659e-07,
|
| 134 |
+
"loss": 4.3,
|
| 135 |
+
"step": 1800
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.03960478592570976,
|
| 139 |
+
"grad_norm": 3.557048797607422,
|
| 140 |
+
"learning_rate": 9.901248078376196e-07,
|
| 141 |
+
"loss": 4.2595,
|
| 142 |
+
"step": 1900
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.04168924834285238,
|
| 146 |
+
"grad_norm": 2.08551025390625,
|
| 147 |
+
"learning_rate": 1.0422366398290733e-06,
|
| 148 |
+
"loss": 4.2167,
|
| 149 |
+
"step": 2000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.043773710759995,
|
| 153 |
+
"grad_norm": 2.8505704402923584,
|
| 154 |
+
"learning_rate": 1.094348471820527e-06,
|
| 155 |
+
"loss": 4.177,
|
| 156 |
+
"step": 2100
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.04585817317713762,
|
| 160 |
+
"grad_norm": 3.39355206489563,
|
| 161 |
+
"learning_rate": 1.1464603038119807e-06,
|
| 162 |
+
"loss": 4.1004,
|
| 163 |
+
"step": 2200
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.04794263559428023,
|
| 167 |
+
"grad_norm": 4.5188093185424805,
|
| 168 |
+
"learning_rate": 1.1985721358034343e-06,
|
| 169 |
+
"loss": 3.9719,
|
| 170 |
+
"step": 2300
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.050027098011422855,
|
| 174 |
+
"grad_norm": 6.547112464904785,
|
| 175 |
+
"learning_rate": 1.2506839677948878e-06,
|
| 176 |
+
"loss": 3.8793,
|
| 177 |
+
"step": 2400
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.052111560428565476,
|
| 181 |
+
"grad_norm": 3.9441330432891846,
|
| 182 |
+
"learning_rate": 1.3027957997863417e-06,
|
| 183 |
+
"loss": 3.7997,
|
| 184 |
+
"step": 2500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.05419602284570809,
|
| 188 |
+
"grad_norm": 18.23339080810547,
|
| 189 |
+
"learning_rate": 1.3549076317777954e-06,
|
| 190 |
+
"loss": 3.6307,
|
| 191 |
+
"step": 2600
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.05628048526285071,
|
| 195 |
+
"grad_norm": 3.295428991317749,
|
| 196 |
+
"learning_rate": 1.4070194637692489e-06,
|
| 197 |
+
"loss": 3.5646,
|
| 198 |
+
"step": 2700
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.05836494767999333,
|
| 202 |
+
"grad_norm": 15.891700744628906,
|
| 203 |
+
"learning_rate": 1.4591312957607026e-06,
|
| 204 |
+
"loss": 3.4813,
|
| 205 |
+
"step": 2800
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.06044941009713595,
|
| 209 |
+
"grad_norm": 5.232209205627441,
|
| 210 |
+
"learning_rate": 1.5112431277521565e-06,
|
| 211 |
+
"loss": 3.389,
|
| 212 |
+
"step": 2900
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.06253387251427857,
|
| 216 |
+
"grad_norm": 9.18393611907959,
|
| 217 |
+
"learning_rate": 1.56335495974361e-06,
|
| 218 |
+
"loss": 3.3027,
|
| 219 |
+
"step": 3000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.06461833493142119,
|
| 223 |
+
"grad_norm": 12.729754447937012,
|
| 224 |
+
"learning_rate": 1.6154667917350636e-06,
|
| 225 |
+
"loss": 3.2259,
|
| 226 |
+
"step": 3100
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.06670279734856381,
|
| 230 |
+
"grad_norm": 4.382412433624268,
|
| 231 |
+
"learning_rate": 1.6675786237265173e-06,
|
| 232 |
+
"loss": 3.1394,
|
| 233 |
+
"step": 3200
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.06878725976570642,
|
| 237 |
+
"grad_norm": 5.850008964538574,
|
| 238 |
+
"learning_rate": 1.7196904557179708e-06,
|
| 239 |
+
"loss": 3.093,
|
| 240 |
+
"step": 3300
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.07087172218284904,
|
| 244 |
+
"grad_norm": 3.021024227142334,
|
| 245 |
+
"learning_rate": 1.7718022877094247e-06,
|
| 246 |
+
"loss": 2.9828,
|
| 247 |
+
"step": 3400
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.07295618459999166,
|
| 251 |
+
"grad_norm": 2.3273675441741943,
|
| 252 |
+
"learning_rate": 1.8239141197008784e-06,
|
| 253 |
+
"loss": 2.921,
|
| 254 |
+
"step": 3500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.07504064701713428,
|
| 258 |
+
"grad_norm": 3.740726947784424,
|
| 259 |
+
"learning_rate": 1.8760259516923318e-06,
|
| 260 |
+
"loss": 2.8224,
|
| 261 |
+
"step": 3600
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.0771251094342769,
|
| 265 |
+
"grad_norm": 2.0490314960479736,
|
| 266 |
+
"learning_rate": 1.9281377836837857e-06,
|
| 267 |
+
"loss": 2.8053,
|
| 268 |
+
"step": 3700
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.07920957185141952,
|
| 272 |
+
"grad_norm": 4.149216651916504,
|
| 273 |
+
"learning_rate": 1.980249615675239e-06,
|
| 274 |
+
"loss": 2.8044,
|
| 275 |
+
"step": 3800
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.08129403426856213,
|
| 279 |
+
"grad_norm": 2.3400914669036865,
|
| 280 |
+
"learning_rate": 2.0323614476666927e-06,
|
| 281 |
+
"loss": 2.7467,
|
| 282 |
+
"step": 3900
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.08337849668570475,
|
| 286 |
+
"grad_norm": 8.31890869140625,
|
| 287 |
+
"learning_rate": 2.0844732796581466e-06,
|
| 288 |
+
"loss": 2.7686,
|
| 289 |
+
"step": 4000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.08546295910284737,
|
| 293 |
+
"grad_norm": 5.363232135772705,
|
| 294 |
+
"learning_rate": 2.1365851116496005e-06,
|
| 295 |
+
"loss": 2.7288,
|
| 296 |
+
"step": 4100
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.08754742151999,
|
| 300 |
+
"grad_norm": 2.7400481700897217,
|
| 301 |
+
"learning_rate": 2.188696943641054e-06,
|
| 302 |
+
"loss": 2.6961,
|
| 303 |
+
"step": 4200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.08963188393713262,
|
| 307 |
+
"grad_norm": 1.5836577415466309,
|
| 308 |
+
"learning_rate": 2.2408087756325074e-06,
|
| 309 |
+
"loss": 2.6586,
|
| 310 |
+
"step": 4300
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.09171634635427524,
|
| 314 |
+
"grad_norm": 1.6861375570297241,
|
| 315 |
+
"learning_rate": 2.2929206076239613e-06,
|
| 316 |
+
"loss": 2.6471,
|
| 317 |
+
"step": 4400
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.09380080877141785,
|
| 321 |
+
"grad_norm": 2.1874806880950928,
|
| 322 |
+
"learning_rate": 2.3450324396154148e-06,
|
| 323 |
+
"loss": 2.64,
|
| 324 |
+
"step": 4500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.09588527118856047,
|
| 328 |
+
"grad_norm": 1.5436290502548218,
|
| 329 |
+
"learning_rate": 2.3971442716068687e-06,
|
| 330 |
+
"loss": 2.6245,
|
| 331 |
+
"step": 4600
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.09796973360570309,
|
| 335 |
+
"grad_norm": 1.6910173892974854,
|
| 336 |
+
"learning_rate": 2.449256103598322e-06,
|
| 337 |
+
"loss": 2.5595,
|
| 338 |
+
"step": 4700
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.10005419602284571,
|
| 342 |
+
"grad_norm": 1.635819673538208,
|
| 343 |
+
"learning_rate": 2.5013679355897756e-06,
|
| 344 |
+
"loss": 2.5527,
|
| 345 |
+
"step": 4800
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.10213865843998833,
|
| 349 |
+
"grad_norm": 1.5254827737808228,
|
| 350 |
+
"learning_rate": 2.5534797675812295e-06,
|
| 351 |
+
"loss": 2.5624,
|
| 352 |
+
"step": 4900
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.10422312085713095,
|
| 356 |
+
"grad_norm": 2.539045810699463,
|
| 357 |
+
"learning_rate": 2.6055915995726834e-06,
|
| 358 |
+
"loss": 2.5594,
|
| 359 |
+
"step": 5000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.10630758327427356,
|
| 363 |
+
"grad_norm": 1.8983627557754517,
|
| 364 |
+
"learning_rate": 2.657703431564137e-06,
|
| 365 |
+
"loss": 2.5547,
|
| 366 |
+
"step": 5100
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.10839204569141618,
|
| 370 |
+
"grad_norm": 1.4213758707046509,
|
| 371 |
+
"learning_rate": 2.709815263555591e-06,
|
| 372 |
+
"loss": 2.5026,
|
| 373 |
+
"step": 5200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.1104765081085588,
|
| 377 |
+
"grad_norm": 1.7480638027191162,
|
| 378 |
+
"learning_rate": 2.761927095547044e-06,
|
| 379 |
+
"loss": 2.514,
|
| 380 |
+
"step": 5300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.11256097052570142,
|
| 384 |
+
"grad_norm": 1.9647337198257446,
|
| 385 |
+
"learning_rate": 2.8140389275384977e-06,
|
| 386 |
+
"loss": 2.5433,
|
| 387 |
+
"step": 5400
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.11464543294284404,
|
| 391 |
+
"grad_norm": 2.1344046592712402,
|
| 392 |
+
"learning_rate": 2.8661507595299516e-06,
|
| 393 |
+
"loss": 2.5185,
|
| 394 |
+
"step": 5500
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.11672989535998667,
|
| 398 |
+
"grad_norm": 2.501701593399048,
|
| 399 |
+
"learning_rate": 2.918262591521405e-06,
|
| 400 |
+
"loss": 2.516,
|
| 401 |
+
"step": 5600
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.11881435777712927,
|
| 405 |
+
"grad_norm": 3.260753631591797,
|
| 406 |
+
"learning_rate": 2.970374423512859e-06,
|
| 407 |
+
"loss": 2.4857,
|
| 408 |
+
"step": 5700
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.1208988201942719,
|
| 412 |
+
"grad_norm": 2.426595687866211,
|
| 413 |
+
"learning_rate": 3.022486255504313e-06,
|
| 414 |
+
"loss": 2.4193,
|
| 415 |
+
"step": 5800
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.12298328261141452,
|
| 419 |
+
"grad_norm": 2.5531206130981445,
|
| 420 |
+
"learning_rate": 3.074598087495766e-06,
|
| 421 |
+
"loss": 2.458,
|
| 422 |
+
"step": 5900
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.12506774502855714,
|
| 426 |
+
"grad_norm": 4.4168243408203125,
|
| 427 |
+
"learning_rate": 3.12670991948722e-06,
|
| 428 |
+
"loss": 2.4794,
|
| 429 |
+
"step": 6000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.12715220744569974,
|
| 433 |
+
"grad_norm": 3.948293924331665,
|
| 434 |
+
"learning_rate": 3.1788217514786733e-06,
|
| 435 |
+
"loss": 2.4691,
|
| 436 |
+
"step": 6100
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.12923666986284238,
|
| 440 |
+
"grad_norm": 1.663045048713684,
|
| 441 |
+
"learning_rate": 3.2309335834701272e-06,
|
| 442 |
+
"loss": 2.4494,
|
| 443 |
+
"step": 6200
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.131321132279985,
|
| 447 |
+
"grad_norm": 3.5404489040374756,
|
| 448 |
+
"learning_rate": 3.283045415461581e-06,
|
| 449 |
+
"loss": 2.428,
|
| 450 |
+
"step": 6300
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.13340559469712762,
|
| 454 |
+
"grad_norm": 1.4624788761138916,
|
| 455 |
+
"learning_rate": 3.3351572474530346e-06,
|
| 456 |
+
"loss": 2.3999,
|
| 457 |
+
"step": 6400
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.13549005711427023,
|
| 461 |
+
"grad_norm": 2.683394432067871,
|
| 462 |
+
"learning_rate": 3.387269079444488e-06,
|
| 463 |
+
"loss": 2.417,
|
| 464 |
+
"step": 6500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.13757451953141284,
|
| 468 |
+
"grad_norm": 2.5286734104156494,
|
| 469 |
+
"learning_rate": 3.4393809114359415e-06,
|
| 470 |
+
"loss": 2.423,
|
| 471 |
+
"step": 6600
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.13965898194855547,
|
| 475 |
+
"grad_norm": 1.595863938331604,
|
| 476 |
+
"learning_rate": 3.4914927434273954e-06,
|
| 477 |
+
"loss": 2.433,
|
| 478 |
+
"step": 6700
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.14174344436569808,
|
| 482 |
+
"grad_norm": 5.466402053833008,
|
| 483 |
+
"learning_rate": 3.5436045754188493e-06,
|
| 484 |
+
"loss": 2.4116,
|
| 485 |
+
"step": 6800
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.14382790678284071,
|
| 489 |
+
"grad_norm": 1.5954893827438354,
|
| 490 |
+
"learning_rate": 3.595716407410303e-06,
|
| 491 |
+
"loss": 2.3939,
|
| 492 |
+
"step": 6900
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.14591236919998332,
|
| 496 |
+
"grad_norm": 2.247056007385254,
|
| 497 |
+
"learning_rate": 3.6478282394017567e-06,
|
| 498 |
+
"loss": 2.3678,
|
| 499 |
+
"step": 7000
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.14799683161712596,
|
| 503 |
+
"grad_norm": 5.489955425262451,
|
| 504 |
+
"learning_rate": 3.6999400713932098e-06,
|
| 505 |
+
"loss": 2.3802,
|
| 506 |
+
"step": 7100
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.15008129403426856,
|
| 510 |
+
"grad_norm": 1.5120584964752197,
|
| 511 |
+
"learning_rate": 3.7520519033846637e-06,
|
| 512 |
+
"loss": 2.3891,
|
| 513 |
+
"step": 7200
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.15216575645141117,
|
| 517 |
+
"grad_norm": 2.228731870651245,
|
| 518 |
+
"learning_rate": 3.8041637353761176e-06,
|
| 519 |
+
"loss": 2.3644,
|
| 520 |
+
"step": 7300
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.1542502188685538,
|
| 524 |
+
"grad_norm": 1.7720115184783936,
|
| 525 |
+
"learning_rate": 3.8562755673675715e-06,
|
| 526 |
+
"loss": 2.3507,
|
| 527 |
+
"step": 7400
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.15633468128569641,
|
| 531 |
+
"grad_norm": 4.0883331298828125,
|
| 532 |
+
"learning_rate": 3.9083873993590245e-06,
|
| 533 |
+
"loss": 2.3572,
|
| 534 |
+
"step": 7500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.15841914370283905,
|
| 538 |
+
"grad_norm": 1.3633663654327393,
|
| 539 |
+
"learning_rate": 3.960499231350478e-06,
|
| 540 |
+
"loss": 2.3387,
|
| 541 |
+
"step": 7600
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.16050360611998166,
|
| 545 |
+
"grad_norm": 3.6679422855377197,
|
| 546 |
+
"learning_rate": 4.0126110633419315e-06,
|
| 547 |
+
"loss": 2.3446,
|
| 548 |
+
"step": 7700
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.16258806853712426,
|
| 552 |
+
"grad_norm": 2.032133102416992,
|
| 553 |
+
"learning_rate": 4.064722895333385e-06,
|
| 554 |
+
"loss": 2.3273,
|
| 555 |
+
"step": 7800
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.1646725309542669,
|
| 559 |
+
"grad_norm": 1.4509003162384033,
|
| 560 |
+
"learning_rate": 4.116834727324839e-06,
|
| 561 |
+
"loss": 2.3072,
|
| 562 |
+
"step": 7900
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.1667569933714095,
|
| 566 |
+
"grad_norm": 4.3479413986206055,
|
| 567 |
+
"learning_rate": 4.168946559316293e-06,
|
| 568 |
+
"loss": 2.3016,
|
| 569 |
+
"step": 8000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.16884145578855214,
|
| 573 |
+
"grad_norm": 12.375444412231445,
|
| 574 |
+
"learning_rate": 4.221058391307747e-06,
|
| 575 |
+
"loss": 2.3145,
|
| 576 |
+
"step": 8100
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.17092591820569475,
|
| 580 |
+
"grad_norm": 1.4382028579711914,
|
| 581 |
+
"learning_rate": 4.273170223299201e-06,
|
| 582 |
+
"loss": 2.274,
|
| 583 |
+
"step": 8200
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.17301038062283736,
|
| 587 |
+
"grad_norm": 7.938540458679199,
|
| 588 |
+
"learning_rate": 4.325282055290654e-06,
|
| 589 |
+
"loss": 2.2873,
|
| 590 |
+
"step": 8300
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.17509484303998,
|
| 594 |
+
"grad_norm": 2.1342923641204834,
|
| 595 |
+
"learning_rate": 4.377393887282108e-06,
|
| 596 |
+
"loss": 2.2906,
|
| 597 |
+
"step": 8400
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.1771793054571226,
|
| 601 |
+
"grad_norm": 1.8758960962295532,
|
| 602 |
+
"learning_rate": 4.429505719273561e-06,
|
| 603 |
+
"loss": 2.2879,
|
| 604 |
+
"step": 8500
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.17926376787426523,
|
| 608 |
+
"grad_norm": 6.642971038818359,
|
| 609 |
+
"learning_rate": 4.481617551265015e-06,
|
| 610 |
+
"loss": 2.2867,
|
| 611 |
+
"step": 8600
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.18134823029140784,
|
| 615 |
+
"grad_norm": 5.171383380889893,
|
| 616 |
+
"learning_rate": 4.533729383256469e-06,
|
| 617 |
+
"loss": 2.2571,
|
| 618 |
+
"step": 8700
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.18343269270855048,
|
| 622 |
+
"grad_norm": 1.664364218711853,
|
| 623 |
+
"learning_rate": 4.585841215247923e-06,
|
| 624 |
+
"loss": 2.2709,
|
| 625 |
+
"step": 8800
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.18551715512569308,
|
| 629 |
+
"grad_norm": 7.0319719314575195,
|
| 630 |
+
"learning_rate": 4.6379530472393765e-06,
|
| 631 |
+
"loss": 2.2643,
|
| 632 |
+
"step": 8900
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.1876016175428357,
|
| 636 |
+
"grad_norm": 2.7922122478485107,
|
| 637 |
+
"learning_rate": 4.6900648792308296e-06,
|
| 638 |
+
"loss": 2.2473,
|
| 639 |
+
"step": 9000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.18968607995997833,
|
| 643 |
+
"grad_norm": 1.8431055545806885,
|
| 644 |
+
"learning_rate": 4.7421767112222835e-06,
|
| 645 |
+
"loss": 2.2499,
|
| 646 |
+
"step": 9100
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.19177054237712093,
|
| 650 |
+
"grad_norm": 3.7758209705352783,
|
| 651 |
+
"learning_rate": 4.794288543213737e-06,
|
| 652 |
+
"loss": 2.2595,
|
| 653 |
+
"step": 9200
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.19385500479426357,
|
| 657 |
+
"grad_norm": 2.5738108158111572,
|
| 658 |
+
"learning_rate": 4.84640037520519e-06,
|
| 659 |
+
"loss": 2.2364,
|
| 660 |
+
"step": 9300
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.19593946721140618,
|
| 664 |
+
"grad_norm": 9.412129402160645,
|
| 665 |
+
"learning_rate": 4.898512207196644e-06,
|
| 666 |
+
"loss": 2.2371,
|
| 667 |
+
"step": 9400
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.19802392962854878,
|
| 671 |
+
"grad_norm": 1.357433795928955,
|
| 672 |
+
"learning_rate": 4.950624039188098e-06,
|
| 673 |
+
"loss": 2.2418,
|
| 674 |
+
"step": 9500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.20010839204569142,
|
| 678 |
+
"grad_norm": 9.067081451416016,
|
| 679 |
+
"learning_rate": 5.002735871179551e-06,
|
| 680 |
+
"loss": 2.2445,
|
| 681 |
+
"step": 9600
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.20219285446283403,
|
| 685 |
+
"grad_norm": 2.770761728286743,
|
| 686 |
+
"learning_rate": 5.054847703171005e-06,
|
| 687 |
+
"loss": 2.2224,
|
| 688 |
+
"step": 9700
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.20427731687997666,
|
| 692 |
+
"grad_norm": 1.7720422744750977,
|
| 693 |
+
"learning_rate": 5.106959535162459e-06,
|
| 694 |
+
"loss": 2.2507,
|
| 695 |
+
"step": 9800
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.20636177929711927,
|
| 699 |
+
"grad_norm": 5.686391830444336,
|
| 700 |
+
"learning_rate": 5.159071367153913e-06,
|
| 701 |
+
"loss": 2.233,
|
| 702 |
+
"step": 9900
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.2084462417142619,
|
| 706 |
+
"grad_norm": 6.246903419494629,
|
| 707 |
+
"learning_rate": 5.211183199145367e-06,
|
| 708 |
+
"loss": 2.2266,
|
| 709 |
+
"step": 10000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.2105307041314045,
|
| 713 |
+
"grad_norm": 7.122190475463867,
|
| 714 |
+
"learning_rate": 5.26329503113682e-06,
|
| 715 |
+
"loss": 2.239,
|
| 716 |
+
"step": 10100
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.21261516654854712,
|
| 720 |
+
"grad_norm": 2.073593854904175,
|
| 721 |
+
"learning_rate": 5.315406863128274e-06,
|
| 722 |
+
"loss": 2.2405,
|
| 723 |
+
"step": 10200
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.21469962896568975,
|
| 727 |
+
"grad_norm": 2.734534740447998,
|
| 728 |
+
"learning_rate": 5.367518695119728e-06,
|
| 729 |
+
"loss": 2.2019,
|
| 730 |
+
"step": 10300
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.21678409138283236,
|
| 734 |
+
"grad_norm": 3.9620697498321533,
|
| 735 |
+
"learning_rate": 5.419630527111182e-06,
|
| 736 |
+
"loss": 2.2132,
|
| 737 |
+
"step": 10400
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.218868553799975,
|
| 741 |
+
"grad_norm": 4.667324542999268,
|
| 742 |
+
"learning_rate": 5.471742359102634e-06,
|
| 743 |
+
"loss": 2.2028,
|
| 744 |
+
"step": 10500
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.2209530162171176,
|
| 748 |
+
"grad_norm": 2.702505111694336,
|
| 749 |
+
"learning_rate": 5.523854191094088e-06,
|
| 750 |
+
"loss": 2.2188,
|
| 751 |
+
"step": 10600
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.2230374786342602,
|
| 755 |
+
"grad_norm": 2.8783977031707764,
|
| 756 |
+
"learning_rate": 5.575966023085542e-06,
|
| 757 |
+
"loss": 2.2058,
|
| 758 |
+
"step": 10700
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.22512194105140285,
|
| 762 |
+
"grad_norm": 4.026909351348877,
|
| 763 |
+
"learning_rate": 5.6280778550769955e-06,
|
| 764 |
+
"loss": 2.2147,
|
| 765 |
+
"step": 10800
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.22720640346854545,
|
| 769 |
+
"grad_norm": 2.730255126953125,
|
| 770 |
+
"learning_rate": 5.680189687068449e-06,
|
| 771 |
+
"loss": 2.203,
|
| 772 |
+
"step": 10900
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.2292908658856881,
|
| 776 |
+
"grad_norm": 1.5225945711135864,
|
| 777 |
+
"learning_rate": 5.732301519059903e-06,
|
| 778 |
+
"loss": 2.1795,
|
| 779 |
+
"step": 11000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.2313753283028307,
|
| 783 |
+
"grad_norm": 2.083085775375366,
|
| 784 |
+
"learning_rate": 5.784413351051356e-06,
|
| 785 |
+
"loss": 2.1758,
|
| 786 |
+
"step": 11100
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.23345979071997333,
|
| 790 |
+
"grad_norm": 3.0502681732177734,
|
| 791 |
+
"learning_rate": 5.83652518304281e-06,
|
| 792 |
+
"loss": 2.1901,
|
| 793 |
+
"step": 11200
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.23554425313711594,
|
| 797 |
+
"grad_norm": 1.8443273305892944,
|
| 798 |
+
"learning_rate": 5.888637015034264e-06,
|
| 799 |
+
"loss": 2.1827,
|
| 800 |
+
"step": 11300
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.23762871555425855,
|
| 804 |
+
"grad_norm": 2.201040506362915,
|
| 805 |
+
"learning_rate": 5.940748847025718e-06,
|
| 806 |
+
"loss": 2.1681,
|
| 807 |
+
"step": 11400
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.23971317797140118,
|
| 811 |
+
"grad_norm": 4.19305419921875,
|
| 812 |
+
"learning_rate": 5.992860679017172e-06,
|
| 813 |
+
"loss": 2.1854,
|
| 814 |
+
"step": 11500
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.2417976403885438,
|
| 818 |
+
"grad_norm": 1.3625741004943848,
|
| 819 |
+
"learning_rate": 6.044972511008626e-06,
|
| 820 |
+
"loss": 2.1777,
|
| 821 |
+
"step": 11600
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.24388210280568642,
|
| 825 |
+
"grad_norm": 2.7626659870147705,
|
| 826 |
+
"learning_rate": 6.097084343000078e-06,
|
| 827 |
+
"loss": 2.1823,
|
| 828 |
+
"step": 11700
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.24596656522282903,
|
| 832 |
+
"grad_norm": 5.543747901916504,
|
| 833 |
+
"learning_rate": 6.149196174991532e-06,
|
| 834 |
+
"loss": 2.1628,
|
| 835 |
+
"step": 11800
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.24805102763997164,
|
| 839 |
+
"grad_norm": 1.4751501083374023,
|
| 840 |
+
"learning_rate": 6.201308006982986e-06,
|
| 841 |
+
"loss": 2.162,
|
| 842 |
+
"step": 11900
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.2501354900571143,
|
| 846 |
+
"grad_norm": 2.3164498805999756,
|
| 847 |
+
"learning_rate": 6.25341983897444e-06,
|
| 848 |
+
"loss": 2.1571,
|
| 849 |
+
"step": 12000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.2522199524742569,
|
| 853 |
+
"grad_norm": 1.5216565132141113,
|
| 854 |
+
"learning_rate": 6.305531670965893e-06,
|
| 855 |
+
"loss": 2.1822,
|
| 856 |
+
"step": 12100
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.2543044148913995,
|
| 860 |
+
"grad_norm": 2.282949924468994,
|
| 861 |
+
"learning_rate": 6.357643502957347e-06,
|
| 862 |
+
"loss": 2.1718,
|
| 863 |
+
"step": 12200
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.25638887730854215,
|
| 867 |
+
"grad_norm": 1.349082112312317,
|
| 868 |
+
"learning_rate": 6.4097553349488006e-06,
|
| 869 |
+
"loss": 2.1843,
|
| 870 |
+
"step": 12300
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.25847333972568476,
|
| 874 |
+
"grad_norm": 3.20103120803833,
|
| 875 |
+
"learning_rate": 6.4618671669402545e-06,
|
| 876 |
+
"loss": 2.1698,
|
| 877 |
+
"step": 12400
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.26055780214282737,
|
| 881 |
+
"grad_norm": 1.9603382349014282,
|
| 882 |
+
"learning_rate": 6.513978998931708e-06,
|
| 883 |
+
"loss": 2.1502,
|
| 884 |
+
"step": 12500
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.26264226455997,
|
| 888 |
+
"grad_norm": 3.995859384536743,
|
| 889 |
+
"learning_rate": 6.566090830923162e-06,
|
| 890 |
+
"loss": 2.1536,
|
| 891 |
+
"step": 12600
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.2647267269771126,
|
| 895 |
+
"grad_norm": 4.198755264282227,
|
| 896 |
+
"learning_rate": 6.618202662914615e-06,
|
| 897 |
+
"loss": 2.1655,
|
| 898 |
+
"step": 12700
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.26681118939425524,
|
| 902 |
+
"grad_norm": 9.24932861328125,
|
| 903 |
+
"learning_rate": 6.670314494906069e-06,
|
| 904 |
+
"loss": 2.1528,
|
| 905 |
+
"step": 12800
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.26889565181139785,
|
| 909 |
+
"grad_norm": 2.8210318088531494,
|
| 910 |
+
"learning_rate": 6.722426326897523e-06,
|
| 911 |
+
"loss": 2.1388,
|
| 912 |
+
"step": 12900
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.27098011422854046,
|
| 916 |
+
"grad_norm": 1.819380760192871,
|
| 917 |
+
"learning_rate": 6.774538158888976e-06,
|
| 918 |
+
"loss": 2.1477,
|
| 919 |
+
"step": 13000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.27306457664568307,
|
| 923 |
+
"grad_norm": 2.15307354927063,
|
| 924 |
+
"learning_rate": 6.826649990880429e-06,
|
| 925 |
+
"loss": 2.1549,
|
| 926 |
+
"step": 13100
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.2751490390628257,
|
| 930 |
+
"grad_norm": 3.6344521045684814,
|
| 931 |
+
"learning_rate": 6.878761822871883e-06,
|
| 932 |
+
"loss": 2.1478,
|
| 933 |
+
"step": 13200
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.27723350147996834,
|
| 937 |
+
"grad_norm": 3.5121541023254395,
|
| 938 |
+
"learning_rate": 6.930873654863337e-06,
|
| 939 |
+
"loss": 2.1366,
|
| 940 |
+
"step": 13300
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.27931796389711094,
|
| 944 |
+
"grad_norm": 1.6062060594558716,
|
| 945 |
+
"learning_rate": 6.982985486854791e-06,
|
| 946 |
+
"loss": 2.1338,
|
| 947 |
+
"step": 13400
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.28140242631425355,
|
| 951 |
+
"grad_norm": 3.3421785831451416,
|
| 952 |
+
"learning_rate": 7.035097318846245e-06,
|
| 953 |
+
"loss": 2.1485,
|
| 954 |
+
"step": 13500
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.28348688873139616,
|
| 958 |
+
"grad_norm": 1.5233893394470215,
|
| 959 |
+
"learning_rate": 7.087209150837699e-06,
|
| 960 |
+
"loss": 2.1412,
|
| 961 |
+
"step": 13600
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.28557135114853877,
|
| 965 |
+
"grad_norm": 12.393982887268066,
|
| 966 |
+
"learning_rate": 7.139320982829152e-06,
|
| 967 |
+
"loss": 2.1399,
|
| 968 |
+
"step": 13700
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.28765581356568143,
|
| 972 |
+
"grad_norm": 2.9735476970672607,
|
| 973 |
+
"learning_rate": 7.191432814820606e-06,
|
| 974 |
+
"loss": 2.143,
|
| 975 |
+
"step": 13800
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.28974027598282404,
|
| 979 |
+
"grad_norm": 2.875746965408325,
|
| 980 |
+
"learning_rate": 7.2435446468120595e-06,
|
| 981 |
+
"loss": 2.1396,
|
| 982 |
+
"step": 13900
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.29182473839996664,
|
| 986 |
+
"grad_norm": 1.34048330783844,
|
| 987 |
+
"learning_rate": 7.2956564788035134e-06,
|
| 988 |
+
"loss": 2.1132,
|
| 989 |
+
"step": 14000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.29390920081710925,
|
| 993 |
+
"grad_norm": 1.4782146215438843,
|
| 994 |
+
"learning_rate": 7.347768310794967e-06,
|
| 995 |
+
"loss": 2.1326,
|
| 996 |
+
"step": 14100
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.2959936632342519,
|
| 1000 |
+
"grad_norm": 2.069708824157715,
|
| 1001 |
+
"learning_rate": 7.3998801427864195e-06,
|
| 1002 |
+
"loss": 2.1433,
|
| 1003 |
+
"step": 14200
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.2980781256513945,
|
| 1007 |
+
"grad_norm": 1.770173192024231,
|
| 1008 |
+
"learning_rate": 7.451991974777873e-06,
|
| 1009 |
+
"loss": 2.1154,
|
| 1010 |
+
"step": 14300
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.30016258806853713,
|
| 1014 |
+
"grad_norm": 2.3121659755706787,
|
| 1015 |
+
"learning_rate": 7.504103806769327e-06,
|
| 1016 |
+
"loss": 2.1279,
|
| 1017 |
+
"step": 14400
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.30224705048567974,
|
| 1021 |
+
"grad_norm": 1.7393426895141602,
|
| 1022 |
+
"learning_rate": 7.556215638760781e-06,
|
| 1023 |
+
"loss": 2.1328,
|
| 1024 |
+
"step": 14500
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.30433151290282234,
|
| 1028 |
+
"grad_norm": 1.6336350440979004,
|
| 1029 |
+
"learning_rate": 7.608327470752235e-06,
|
| 1030 |
+
"loss": 2.0962,
|
| 1031 |
+
"step": 14600
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.306415975319965,
|
| 1035 |
+
"grad_norm": 1.4256808757781982,
|
| 1036 |
+
"learning_rate": 7.660439302743689e-06,
|
| 1037 |
+
"loss": 2.1283,
|
| 1038 |
+
"step": 14700
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.3085004377371076,
|
| 1042 |
+
"grad_norm": 1.3236781358718872,
|
| 1043 |
+
"learning_rate": 7.712551134735143e-06,
|
| 1044 |
+
"loss": 2.1225,
|
| 1045 |
+
"step": 14800
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.3105849001542502,
|
| 1049 |
+
"grad_norm": 15.130003929138184,
|
| 1050 |
+
"learning_rate": 7.764662966726597e-06,
|
| 1051 |
+
"loss": 2.1256,
|
| 1052 |
+
"step": 14900
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.31266936257139283,
|
| 1056 |
+
"grad_norm": 3.1798925399780273,
|
| 1057 |
+
"learning_rate": 7.816774798718049e-06,
|
| 1058 |
+
"loss": 2.1199,
|
| 1059 |
+
"step": 15000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.31475382498853544,
|
| 1063 |
+
"grad_norm": 3.058833599090576,
|
| 1064 |
+
"learning_rate": 7.868886630709503e-06,
|
| 1065 |
+
"loss": 2.1067,
|
| 1066 |
+
"step": 15100
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.3168382874056781,
|
| 1070 |
+
"grad_norm": 2.1483590602874756,
|
| 1071 |
+
"learning_rate": 7.920998462700957e-06,
|
| 1072 |
+
"loss": 2.0995,
|
| 1073 |
+
"step": 15200
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.3189227498228207,
|
| 1077 |
+
"grad_norm": 1.3835676908493042,
|
| 1078 |
+
"learning_rate": 7.97311029469241e-06,
|
| 1079 |
+
"loss": 2.1013,
|
| 1080 |
+
"step": 15300
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.3210072122399633,
|
| 1084 |
+
"grad_norm": 1.9250987768173218,
|
| 1085 |
+
"learning_rate": 8.025222126683863e-06,
|
| 1086 |
+
"loss": 2.1124,
|
| 1087 |
+
"step": 15400
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.3230916746571059,
|
| 1091 |
+
"grad_norm": 10.668524742126465,
|
| 1092 |
+
"learning_rate": 8.077333958675317e-06,
|
| 1093 |
+
"loss": 2.0705,
|
| 1094 |
+
"step": 15500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.32517613707424853,
|
| 1098 |
+
"grad_norm": 1.4059637784957886,
|
| 1099 |
+
"learning_rate": 8.12944579066677e-06,
|
| 1100 |
+
"loss": 2.1161,
|
| 1101 |
+
"step": 15600
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.3272605994913912,
|
| 1105 |
+
"grad_norm": 1.4063282012939453,
|
| 1106 |
+
"learning_rate": 8.181557622658225e-06,
|
| 1107 |
+
"loss": 2.1158,
|
| 1108 |
+
"step": 15700
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.3293450619085338,
|
| 1112 |
+
"grad_norm": 1.4921411275863647,
|
| 1113 |
+
"learning_rate": 8.233669454649678e-06,
|
| 1114 |
+
"loss": 2.0935,
|
| 1115 |
+
"step": 15800
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.3314295243256764,
|
| 1119 |
+
"grad_norm": 1.4278446435928345,
|
| 1120 |
+
"learning_rate": 8.285781286641132e-06,
|
| 1121 |
+
"loss": 2.0921,
|
| 1122 |
+
"step": 15900
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.333513986742819,
|
| 1126 |
+
"grad_norm": 1.509752631187439,
|
| 1127 |
+
"learning_rate": 8.337893118632586e-06,
|
| 1128 |
+
"loss": 2.0946,
|
| 1129 |
+
"step": 16000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.3355984491599616,
|
| 1133 |
+
"grad_norm": 3.8793394565582275,
|
| 1134 |
+
"learning_rate": 8.39000495062404e-06,
|
| 1135 |
+
"loss": 2.0952,
|
| 1136 |
+
"step": 16100
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.3376829115771043,
|
| 1140 |
+
"grad_norm": 2.2685413360595703,
|
| 1141 |
+
"learning_rate": 8.442116782615494e-06,
|
| 1142 |
+
"loss": 2.0802,
|
| 1143 |
+
"step": 16200
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.3397673739942469,
|
| 1147 |
+
"grad_norm": 1.4082143306732178,
|
| 1148 |
+
"learning_rate": 8.494228614606948e-06,
|
| 1149 |
+
"loss": 2.0913,
|
| 1150 |
+
"step": 16300
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.3418518364113895,
|
| 1154 |
+
"grad_norm": 2.058350086212158,
|
| 1155 |
+
"learning_rate": 8.546340446598402e-06,
|
| 1156 |
+
"loss": 2.1075,
|
| 1157 |
+
"step": 16400
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.3439362988285321,
|
| 1161 |
+
"grad_norm": 1.5691351890563965,
|
| 1162 |
+
"learning_rate": 8.598452278589856e-06,
|
| 1163 |
+
"loss": 2.0984,
|
| 1164 |
+
"step": 16500
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.3460207612456747,
|
| 1168 |
+
"grad_norm": 2.385117769241333,
|
| 1169 |
+
"learning_rate": 8.650564110581308e-06,
|
| 1170 |
+
"loss": 2.0793,
|
| 1171 |
+
"step": 16600
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.3481052236628174,
|
| 1175 |
+
"grad_norm": 2.0215940475463867,
|
| 1176 |
+
"learning_rate": 8.702675942572762e-06,
|
| 1177 |
+
"loss": 2.0739,
|
| 1178 |
+
"step": 16700
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.35018968607996,
|
| 1182 |
+
"grad_norm": 1.5603607892990112,
|
| 1183 |
+
"learning_rate": 8.754787774564216e-06,
|
| 1184 |
+
"loss": 2.0733,
|
| 1185 |
+
"step": 16800
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.3522741484971026,
|
| 1189 |
+
"grad_norm": 1.732659101486206,
|
| 1190 |
+
"learning_rate": 8.80689960655567e-06,
|
| 1191 |
+
"loss": 2.0928,
|
| 1192 |
+
"step": 16900
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.3543586109142452,
|
| 1196 |
+
"grad_norm": 18.339719772338867,
|
| 1197 |
+
"learning_rate": 8.859011438547122e-06,
|
| 1198 |
+
"loss": 2.0802,
|
| 1199 |
+
"step": 17000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.35644307333138786,
|
| 1203 |
+
"grad_norm": 1.269181251525879,
|
| 1204 |
+
"learning_rate": 8.911123270538576e-06,
|
| 1205 |
+
"loss": 2.063,
|
| 1206 |
+
"step": 17100
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.35852753574853047,
|
| 1210 |
+
"grad_norm": 1.4720960855484009,
|
| 1211 |
+
"learning_rate": 8.96323510253003e-06,
|
| 1212 |
+
"loss": 2.0952,
|
| 1213 |
+
"step": 17200
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.3606119981656731,
|
| 1217 |
+
"grad_norm": 1.8136069774627686,
|
| 1218 |
+
"learning_rate": 9.015346934521484e-06,
|
| 1219 |
+
"loss": 2.0834,
|
| 1220 |
+
"step": 17300
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.3626964605828157,
|
| 1224 |
+
"grad_norm": 1.4650546312332153,
|
| 1225 |
+
"learning_rate": 9.067458766512937e-06,
|
| 1226 |
+
"loss": 2.0775,
|
| 1227 |
+
"step": 17400
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.3647809229999583,
|
| 1231 |
+
"grad_norm": 1.538031816482544,
|
| 1232 |
+
"learning_rate": 9.119570598504391e-06,
|
| 1233 |
+
"loss": 2.094,
|
| 1234 |
+
"step": 17500
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.36686538541710095,
|
| 1238 |
+
"grad_norm": 1.2462719678878784,
|
| 1239 |
+
"learning_rate": 9.171682430495845e-06,
|
| 1240 |
+
"loss": 2.072,
|
| 1241 |
+
"step": 17600
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.36894984783424356,
|
| 1245 |
+
"grad_norm": 1.4545553922653198,
|
| 1246 |
+
"learning_rate": 9.223794262487299e-06,
|
| 1247 |
+
"loss": 2.0787,
|
| 1248 |
+
"step": 17700
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.37103431025138617,
|
| 1252 |
+
"grad_norm": 1.3159983158111572,
|
| 1253 |
+
"learning_rate": 9.275906094478753e-06,
|
| 1254 |
+
"loss": 2.0681,
|
| 1255 |
+
"step": 17800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.3731187726685288,
|
| 1259 |
+
"grad_norm": 4.405596733093262,
|
| 1260 |
+
"learning_rate": 9.328017926470205e-06,
|
| 1261 |
+
"loss": 2.0666,
|
| 1262 |
+
"step": 17900
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.3752032350856714,
|
| 1266 |
+
"grad_norm": 1.6247169971466064,
|
| 1267 |
+
"learning_rate": 9.380129758461659e-06,
|
| 1268 |
+
"loss": 2.0682,
|
| 1269 |
+
"step": 18000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.37728769750281405,
|
| 1273 |
+
"grad_norm": 1.2033179998397827,
|
| 1274 |
+
"learning_rate": 9.432241590453113e-06,
|
| 1275 |
+
"loss": 2.0714,
|
| 1276 |
+
"step": 18100
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.37937215991995665,
|
| 1280 |
+
"grad_norm": 1.3838162422180176,
|
| 1281 |
+
"learning_rate": 9.484353422444567e-06,
|
| 1282 |
+
"loss": 2.0692,
|
| 1283 |
+
"step": 18200
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.38145662233709926,
|
| 1287 |
+
"grad_norm": 1.642264723777771,
|
| 1288 |
+
"learning_rate": 9.53646525443602e-06,
|
| 1289 |
+
"loss": 2.0674,
|
| 1290 |
+
"step": 18300
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.38354108475424187,
|
| 1294 |
+
"grad_norm": 1.152984619140625,
|
| 1295 |
+
"learning_rate": 9.588577086427475e-06,
|
| 1296 |
+
"loss": 2.0644,
|
| 1297 |
+
"step": 18400
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.3856255471713845,
|
| 1301 |
+
"grad_norm": 3.9230525493621826,
|
| 1302 |
+
"learning_rate": 9.640688918418929e-06,
|
| 1303 |
+
"loss": 2.0642,
|
| 1304 |
+
"step": 18500
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.38771000958852714,
|
| 1308 |
+
"grad_norm": 1.4267147779464722,
|
| 1309 |
+
"learning_rate": 9.69280075041038e-06,
|
| 1310 |
+
"loss": 2.0842,
|
| 1311 |
+
"step": 18600
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.38979447200566975,
|
| 1315 |
+
"grad_norm": 2.8921425342559814,
|
| 1316 |
+
"learning_rate": 9.744912582401835e-06,
|
| 1317 |
+
"loss": 2.0604,
|
| 1318 |
+
"step": 18700
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.39187893442281235,
|
| 1322 |
+
"grad_norm": 1.3720977306365967,
|
| 1323 |
+
"learning_rate": 9.797024414393289e-06,
|
| 1324 |
+
"loss": 2.0512,
|
| 1325 |
+
"step": 18800
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.39396339683995496,
|
| 1329 |
+
"grad_norm": 2.229217052459717,
|
| 1330 |
+
"learning_rate": 9.849136246384743e-06,
|
| 1331 |
+
"loss": 2.0377,
|
| 1332 |
+
"step": 18900
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.39604785925709757,
|
| 1336 |
+
"grad_norm": 1.2066770792007446,
|
| 1337 |
+
"learning_rate": 9.901248078376196e-06,
|
| 1338 |
+
"loss": 2.0627,
|
| 1339 |
+
"step": 19000
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.39813232167424023,
|
| 1343 |
+
"grad_norm": 1.3302040100097656,
|
| 1344 |
+
"learning_rate": 9.953359910367649e-06,
|
| 1345 |
+
"loss": 2.063,
|
| 1346 |
+
"step": 19100
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.40021678409138284,
|
| 1350 |
+
"grad_norm": 2.0192320346832275,
|
| 1351 |
+
"learning_rate": 1.0005471742359103e-05,
|
| 1352 |
+
"loss": 2.0521,
|
| 1353 |
+
"step": 19200
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.40230124650852545,
|
| 1357 |
+
"grad_norm": 1.3492109775543213,
|
| 1358 |
+
"learning_rate": 1.0057583574350558e-05,
|
| 1359 |
+
"loss": 2.0433,
|
| 1360 |
+
"step": 19300
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.40438570892566805,
|
| 1364 |
+
"grad_norm": 1.3807893991470337,
|
| 1365 |
+
"learning_rate": 1.010969540634201e-05,
|
| 1366 |
+
"loss": 2.0624,
|
| 1367 |
+
"step": 19400
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.4064701713428107,
|
| 1371 |
+
"grad_norm": 1.525590419769287,
|
| 1372 |
+
"learning_rate": 1.0161807238333466e-05,
|
| 1373 |
+
"loss": 2.0475,
|
| 1374 |
+
"step": 19500
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.4085546337599533,
|
| 1378 |
+
"grad_norm": 1.2582364082336426,
|
| 1379 |
+
"learning_rate": 1.0213919070324918e-05,
|
| 1380 |
+
"loss": 2.0339,
|
| 1381 |
+
"step": 19600
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.41063909617709593,
|
| 1385 |
+
"grad_norm": 2.245859146118164,
|
| 1386 |
+
"learning_rate": 1.026603090231637e-05,
|
| 1387 |
+
"loss": 2.0401,
|
| 1388 |
+
"step": 19700
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.41272355859423854,
|
| 1392 |
+
"grad_norm": 1.3547762632369995,
|
| 1393 |
+
"learning_rate": 1.0318142734307826e-05,
|
| 1394 |
+
"loss": 2.035,
|
| 1395 |
+
"step": 19800
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.41480802101138115,
|
| 1399 |
+
"grad_norm": 1.5345648527145386,
|
| 1400 |
+
"learning_rate": 1.0370254566299278e-05,
|
| 1401 |
+
"loss": 2.048,
|
| 1402 |
+
"step": 19900
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.4168924834285238,
|
| 1406 |
+
"grad_norm": 9.708149909973145,
|
| 1407 |
+
"learning_rate": 1.0422366398290734e-05,
|
| 1408 |
+
"loss": 2.0277,
|
| 1409 |
+
"step": 20000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.4189769458456664,
|
| 1413 |
+
"grad_norm": 1.277744174003601,
|
| 1414 |
+
"learning_rate": 1.0474478230282186e-05,
|
| 1415 |
+
"loss": 2.046,
|
| 1416 |
+
"step": 20100
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.421061408262809,
|
| 1420 |
+
"grad_norm": 1.1756649017333984,
|
| 1421 |
+
"learning_rate": 1.052659006227364e-05,
|
| 1422 |
+
"loss": 2.0232,
|
| 1423 |
+
"step": 20200
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.42314587067995163,
|
| 1427 |
+
"grad_norm": 1.0774590969085693,
|
| 1428 |
+
"learning_rate": 1.0578701894265094e-05,
|
| 1429 |
+
"loss": 2.0486,
|
| 1430 |
+
"step": 20300
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.42523033309709424,
|
| 1434 |
+
"grad_norm": 1.1391072273254395,
|
| 1435 |
+
"learning_rate": 1.0630813726256548e-05,
|
| 1436 |
+
"loss": 2.0329,
|
| 1437 |
+
"step": 20400
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.4273147955142369,
|
| 1441 |
+
"grad_norm": 1.9197932481765747,
|
| 1442 |
+
"learning_rate": 1.0682925558248001e-05,
|
| 1443 |
+
"loss": 2.0328,
|
| 1444 |
+
"step": 20500
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.4293992579313795,
|
| 1448 |
+
"grad_norm": 1.3254817724227905,
|
| 1449 |
+
"learning_rate": 1.0735037390239455e-05,
|
| 1450 |
+
"loss": 2.0308,
|
| 1451 |
+
"step": 20600
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.4314837203485221,
|
| 1455 |
+
"grad_norm": 4.676088809967041,
|
| 1456 |
+
"learning_rate": 1.0787149222230908e-05,
|
| 1457 |
+
"loss": 2.0502,
|
| 1458 |
+
"step": 20700
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.4335681827656647,
|
| 1462 |
+
"grad_norm": 1.493910551071167,
|
| 1463 |
+
"learning_rate": 1.0839261054222363e-05,
|
| 1464 |
+
"loss": 2.0197,
|
| 1465 |
+
"step": 20800
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.43565264518280733,
|
| 1469 |
+
"grad_norm": 1.8109139204025269,
|
| 1470 |
+
"learning_rate": 1.0891372886213815e-05,
|
| 1471 |
+
"loss": 2.055,
|
| 1472 |
+
"step": 20900
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.43773710759995,
|
| 1476 |
+
"grad_norm": 1.8932719230651855,
|
| 1477 |
+
"learning_rate": 1.0943484718205268e-05,
|
| 1478 |
+
"loss": 2.0294,
|
| 1479 |
+
"step": 21000
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.4398215700170926,
|
| 1483 |
+
"grad_norm": 1.4016826152801514,
|
| 1484 |
+
"learning_rate": 1.0995596550196723e-05,
|
| 1485 |
+
"loss": 2.0341,
|
| 1486 |
+
"step": 21100
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.4419060324342352,
|
| 1490 |
+
"grad_norm": 1.1845966577529907,
|
| 1491 |
+
"learning_rate": 1.1047708382188175e-05,
|
| 1492 |
+
"loss": 2.04,
|
| 1493 |
+
"step": 21200
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.4439904948513778,
|
| 1497 |
+
"grad_norm": 9.983181953430176,
|
| 1498 |
+
"learning_rate": 1.1099820214179631e-05,
|
| 1499 |
+
"loss": 2.0413,
|
| 1500 |
+
"step": 21300
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.4460749572685204,
|
| 1504 |
+
"grad_norm": 1.4713338613510132,
|
| 1505 |
+
"learning_rate": 1.1151932046171083e-05,
|
| 1506 |
+
"loss": 2.0447,
|
| 1507 |
+
"step": 21400
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.4481594196856631,
|
| 1511 |
+
"grad_norm": 1.1357747316360474,
|
| 1512 |
+
"learning_rate": 1.1204043878162539e-05,
|
| 1513 |
+
"loss": 2.008,
|
| 1514 |
+
"step": 21500
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.4502438821028057,
|
| 1518 |
+
"grad_norm": 1.479264497756958,
|
| 1519 |
+
"learning_rate": 1.1256155710153991e-05,
|
| 1520 |
+
"loss": 2.0303,
|
| 1521 |
+
"step": 21600
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.4523283445199483,
|
| 1525 |
+
"grad_norm": 1.3185663223266602,
|
| 1526 |
+
"learning_rate": 1.1308267542145447e-05,
|
| 1527 |
+
"loss": 2.0427,
|
| 1528 |
+
"step": 21700
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.4544128069370909,
|
| 1532 |
+
"grad_norm": 1.1067243814468384,
|
| 1533 |
+
"learning_rate": 1.1360379374136899e-05,
|
| 1534 |
+
"loss": 2.0116,
|
| 1535 |
+
"step": 21800
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.45649726935423357,
|
| 1539 |
+
"grad_norm": 1.9280306100845337,
|
| 1540 |
+
"learning_rate": 1.1412491206128353e-05,
|
| 1541 |
+
"loss": 2.0235,
|
| 1542 |
+
"step": 21900
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.4585817317713762,
|
| 1546 |
+
"grad_norm": 1.7216917276382446,
|
| 1547 |
+
"learning_rate": 1.1464603038119807e-05,
|
| 1548 |
+
"loss": 2.0208,
|
| 1549 |
+
"step": 22000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.4606661941885188,
|
| 1553 |
+
"grad_norm": 1.0976884365081787,
|
| 1554 |
+
"learning_rate": 1.151671487011126e-05,
|
| 1555 |
+
"loss": 2.0185,
|
| 1556 |
+
"step": 22100
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.4627506566056614,
|
| 1560 |
+
"grad_norm": 1.3677102327346802,
|
| 1561 |
+
"learning_rate": 1.1568826702102713e-05,
|
| 1562 |
+
"loss": 2.0132,
|
| 1563 |
+
"step": 22200
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.464835119022804,
|
| 1567 |
+
"grad_norm": 1.1687935590744019,
|
| 1568 |
+
"learning_rate": 1.1620938534094167e-05,
|
| 1569 |
+
"loss": 2.0224,
|
| 1570 |
+
"step": 22300
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 0.46691958143994666,
|
| 1574 |
+
"grad_norm": 2.3308753967285156,
|
| 1575 |
+
"learning_rate": 1.167305036608562e-05,
|
| 1576 |
+
"loss": 2.0277,
|
| 1577 |
+
"step": 22400
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.46900404385708927,
|
| 1581 |
+
"grad_norm": 1.6247777938842773,
|
| 1582 |
+
"learning_rate": 1.1725162198077074e-05,
|
| 1583 |
+
"loss": 2.0265,
|
| 1584 |
+
"step": 22500
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.4710885062742319,
|
| 1588 |
+
"grad_norm": 1.235781192779541,
|
| 1589 |
+
"learning_rate": 1.1777274030068528e-05,
|
| 1590 |
+
"loss": 2.0124,
|
| 1591 |
+
"step": 22600
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.4731729686913745,
|
| 1595 |
+
"grad_norm": 1.9984397888183594,
|
| 1596 |
+
"learning_rate": 1.182938586205998e-05,
|
| 1597 |
+
"loss": 1.9896,
|
| 1598 |
+
"step": 22700
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.4752574311085171,
|
| 1602 |
+
"grad_norm": 1.3742737770080566,
|
| 1603 |
+
"learning_rate": 1.1881497694051436e-05,
|
| 1604 |
+
"loss": 2.0,
|
| 1605 |
+
"step": 22800
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.47734189352565976,
|
| 1609 |
+
"grad_norm": 2.8917479515075684,
|
| 1610 |
+
"learning_rate": 1.1933609526042888e-05,
|
| 1611 |
+
"loss": 2.0025,
|
| 1612 |
+
"step": 22900
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.47942635594280236,
|
| 1616 |
+
"grad_norm": 1.2879999876022339,
|
| 1617 |
+
"learning_rate": 1.1985721358034344e-05,
|
| 1618 |
+
"loss": 2.0312,
|
| 1619 |
+
"step": 23000
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.48151081835994497,
|
| 1623 |
+
"grad_norm": 0.9497489929199219,
|
| 1624 |
+
"learning_rate": 1.2037833190025796e-05,
|
| 1625 |
+
"loss": 2.0129,
|
| 1626 |
+
"step": 23100
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.4835952807770876,
|
| 1630 |
+
"grad_norm": 1.982295274734497,
|
| 1631 |
+
"learning_rate": 1.2089945022017252e-05,
|
| 1632 |
+
"loss": 2.0036,
|
| 1633 |
+
"step": 23200
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.4856797431942302,
|
| 1637 |
+
"grad_norm": 1.4037415981292725,
|
| 1638 |
+
"learning_rate": 1.2142056854008704e-05,
|
| 1639 |
+
"loss": 1.9977,
|
| 1640 |
+
"step": 23300
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.48776420561137285,
|
| 1644 |
+
"grad_norm": 1.1049786806106567,
|
| 1645 |
+
"learning_rate": 1.2194168686000156e-05,
|
| 1646 |
+
"loss": 2.0144,
|
| 1647 |
+
"step": 23400
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.48984866802851545,
|
| 1651 |
+
"grad_norm": 1.9810779094696045,
|
| 1652 |
+
"learning_rate": 1.2246280517991612e-05,
|
| 1653 |
+
"loss": 2.004,
|
| 1654 |
+
"step": 23500
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.49193313044565806,
|
| 1658 |
+
"grad_norm": 1.4957938194274902,
|
| 1659 |
+
"learning_rate": 1.2298392349983064e-05,
|
| 1660 |
+
"loss": 1.9874,
|
| 1661 |
+
"step": 23600
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.49401759286280067,
|
| 1665 |
+
"grad_norm": 1.414520263671875,
|
| 1666 |
+
"learning_rate": 1.235050418197452e-05,
|
| 1667 |
+
"loss": 2.0238,
|
| 1668 |
+
"step": 23700
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.4961020552799433,
|
| 1672 |
+
"grad_norm": 1.2930928468704224,
|
| 1673 |
+
"learning_rate": 1.2402616013965972e-05,
|
| 1674 |
+
"loss": 1.9898,
|
| 1675 |
+
"step": 23800
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.49818651769708594,
|
| 1679 |
+
"grad_norm": 1.3122200965881348,
|
| 1680 |
+
"learning_rate": 1.2454727845957426e-05,
|
| 1681 |
+
"loss": 1.9952,
|
| 1682 |
+
"step": 23900
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.5002709801142285,
|
| 1686 |
+
"grad_norm": 3.088305950164795,
|
| 1687 |
+
"learning_rate": 1.250683967794888e-05,
|
| 1688 |
+
"loss": 1.9919,
|
| 1689 |
+
"step": 24000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.5023554425313712,
|
| 1693 |
+
"grad_norm": 1.2572028636932373,
|
| 1694 |
+
"learning_rate": 1.2558951509940333e-05,
|
| 1695 |
+
"loss": 2.0099,
|
| 1696 |
+
"step": 24100
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 0.5044399049485138,
|
| 1700 |
+
"grad_norm": 1.3861281871795654,
|
| 1701 |
+
"learning_rate": 1.2611063341931786e-05,
|
| 1702 |
+
"loss": 2.0091,
|
| 1703 |
+
"step": 24200
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.5065243673656564,
|
| 1707 |
+
"grad_norm": 3.073563814163208,
|
| 1708 |
+
"learning_rate": 1.2663175173923241e-05,
|
| 1709 |
+
"loss": 1.995,
|
| 1710 |
+
"step": 24300
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 0.508608829782799,
|
| 1714 |
+
"grad_norm": 0.9809319972991943,
|
| 1715 |
+
"learning_rate": 1.2715287005914693e-05,
|
| 1716 |
+
"loss": 2.013,
|
| 1717 |
+
"step": 24400
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 0.5106932921999416,
|
| 1721 |
+
"grad_norm": 1.243375301361084,
|
| 1722 |
+
"learning_rate": 1.2767398837906149e-05,
|
| 1723 |
+
"loss": 1.9997,
|
| 1724 |
+
"step": 24500
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.5127777546170843,
|
| 1728 |
+
"grad_norm": 1.1517629623413086,
|
| 1729 |
+
"learning_rate": 1.2819510669897601e-05,
|
| 1730 |
+
"loss": 1.988,
|
| 1731 |
+
"step": 24600
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.5148622170342269,
|
| 1735 |
+
"grad_norm": 1.890515685081482,
|
| 1736 |
+
"learning_rate": 1.2871622501889053e-05,
|
| 1737 |
+
"loss": 1.9872,
|
| 1738 |
+
"step": 24700
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.5169466794513695,
|
| 1742 |
+
"grad_norm": 1.2368491888046265,
|
| 1743 |
+
"learning_rate": 1.2923734333880509e-05,
|
| 1744 |
+
"loss": 1.9788,
|
| 1745 |
+
"step": 24800
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.5190311418685121,
|
| 1749 |
+
"grad_norm": 1.298607587814331,
|
| 1750 |
+
"learning_rate": 1.2975846165871961e-05,
|
| 1751 |
+
"loss": 1.9849,
|
| 1752 |
+
"step": 24900
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.5211156042856547,
|
| 1756 |
+
"grad_norm": 1.2520874738693237,
|
| 1757 |
+
"learning_rate": 1.3027957997863417e-05,
|
| 1758 |
+
"loss": 2.0153,
|
| 1759 |
+
"step": 25000
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.5232000667027974,
|
| 1763 |
+
"grad_norm": 1.6040500402450562,
|
| 1764 |
+
"learning_rate": 1.3080069829854869e-05,
|
| 1765 |
+
"loss": 1.9786,
|
| 1766 |
+
"step": 25100
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.52528452911994,
|
| 1770 |
+
"grad_norm": 1.2517794370651245,
|
| 1771 |
+
"learning_rate": 1.3132181661846325e-05,
|
| 1772 |
+
"loss": 1.9779,
|
| 1773 |
+
"step": 25200
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.5273689915370826,
|
| 1777 |
+
"grad_norm": 1.0364949703216553,
|
| 1778 |
+
"learning_rate": 1.3184293493837777e-05,
|
| 1779 |
+
"loss": 2.001,
|
| 1780 |
+
"step": 25300
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 0.5294534539542252,
|
| 1784 |
+
"grad_norm": 1.2333202362060547,
|
| 1785 |
+
"learning_rate": 1.323640532582923e-05,
|
| 1786 |
+
"loss": 2.0127,
|
| 1787 |
+
"step": 25400
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.5315379163713678,
|
| 1791 |
+
"grad_norm": 1.1749173402786255,
|
| 1792 |
+
"learning_rate": 1.3288517157820685e-05,
|
| 1793 |
+
"loss": 1.9699,
|
| 1794 |
+
"step": 25500
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 0.5336223787885105,
|
| 1798 |
+
"grad_norm": 1.2822164297103882,
|
| 1799 |
+
"learning_rate": 1.3340628989812138e-05,
|
| 1800 |
+
"loss": 1.9742,
|
| 1801 |
+
"step": 25600
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.535706841205653,
|
| 1805 |
+
"grad_norm": 12.977947235107422,
|
| 1806 |
+
"learning_rate": 1.3392740821803592e-05,
|
| 1807 |
+
"loss": 1.9848,
|
| 1808 |
+
"step": 25700
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.5377913036227957,
|
| 1812 |
+
"grad_norm": 1.4702880382537842,
|
| 1813 |
+
"learning_rate": 1.3444852653795046e-05,
|
| 1814 |
+
"loss": 1.9828,
|
| 1815 |
+
"step": 25800
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.5398757660399383,
|
| 1819 |
+
"grad_norm": 1.9451568126678467,
|
| 1820 |
+
"learning_rate": 1.3496964485786498e-05,
|
| 1821 |
+
"loss": 1.9818,
|
| 1822 |
+
"step": 25900
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 0.5419602284570809,
|
| 1826 |
+
"grad_norm": 1.347174882888794,
|
| 1827 |
+
"learning_rate": 1.3549076317777952e-05,
|
| 1828 |
+
"loss": 1.9821,
|
| 1829 |
+
"step": 26000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.5440446908742236,
|
| 1833 |
+
"grad_norm": 1.4248614311218262,
|
| 1834 |
+
"learning_rate": 1.3601188149769406e-05,
|
| 1835 |
+
"loss": 1.9753,
|
| 1836 |
+
"step": 26100
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 0.5461291532913661,
|
| 1840 |
+
"grad_norm": 1.8619277477264404,
|
| 1841 |
+
"learning_rate": 1.3653299981760858e-05,
|
| 1842 |
+
"loss": 1.9882,
|
| 1843 |
+
"step": 26200
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 0.5482136157085088,
|
| 1847 |
+
"grad_norm": 1.6204618215560913,
|
| 1848 |
+
"learning_rate": 1.3705411813752314e-05,
|
| 1849 |
+
"loss": 1.9807,
|
| 1850 |
+
"step": 26300
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.5502980781256513,
|
| 1854 |
+
"grad_norm": 1.8339996337890625,
|
| 1855 |
+
"learning_rate": 1.3757523645743766e-05,
|
| 1856 |
+
"loss": 1.9563,
|
| 1857 |
+
"step": 26400
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 0.552382540542794,
|
| 1861 |
+
"grad_norm": 1.0893349647521973,
|
| 1862 |
+
"learning_rate": 1.3809635477735222e-05,
|
| 1863 |
+
"loss": 1.9731,
|
| 1864 |
+
"step": 26500
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 0.5544670029599367,
|
| 1868 |
+
"grad_norm": 1.1871613264083862,
|
| 1869 |
+
"learning_rate": 1.3861747309726674e-05,
|
| 1870 |
+
"loss": 1.9645,
|
| 1871 |
+
"step": 26600
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 0.5565514653770792,
|
| 1875 |
+
"grad_norm": 1.064344882965088,
|
| 1876 |
+
"learning_rate": 1.391385914171813e-05,
|
| 1877 |
+
"loss": 1.9775,
|
| 1878 |
+
"step": 26700
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 0.5586359277942219,
|
| 1882 |
+
"grad_norm": 0.9557547569274902,
|
| 1883 |
+
"learning_rate": 1.3965970973709582e-05,
|
| 1884 |
+
"loss": 1.9658,
|
| 1885 |
+
"step": 26800
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 0.5607203902113644,
|
| 1889 |
+
"grad_norm": 1.0063979625701904,
|
| 1890 |
+
"learning_rate": 1.4018082805701037e-05,
|
| 1891 |
+
"loss": 1.9578,
|
| 1892 |
+
"step": 26900
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.5628048526285071,
|
| 1896 |
+
"grad_norm": 1.4233479499816895,
|
| 1897 |
+
"learning_rate": 1.407019463769249e-05,
|
| 1898 |
+
"loss": 1.971,
|
| 1899 |
+
"step": 27000
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.5648893150456498,
|
| 1903 |
+
"grad_norm": 1.1650019884109497,
|
| 1904 |
+
"learning_rate": 1.4122306469683942e-05,
|
| 1905 |
+
"loss": 1.9632,
|
| 1906 |
+
"step": 27100
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 0.5669737774627923,
|
| 1910 |
+
"grad_norm": 1.025070309638977,
|
| 1911 |
+
"learning_rate": 1.4174418301675397e-05,
|
| 1912 |
+
"loss": 1.9593,
|
| 1913 |
+
"step": 27200
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 0.569058239879935,
|
| 1917 |
+
"grad_norm": 0.8504909873008728,
|
| 1918 |
+
"learning_rate": 1.422653013366685e-05,
|
| 1919 |
+
"loss": 1.9738,
|
| 1920 |
+
"step": 27300
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 0.5711427022970775,
|
| 1924 |
+
"grad_norm": 1.2217037677764893,
|
| 1925 |
+
"learning_rate": 1.4278641965658303e-05,
|
| 1926 |
+
"loss": 1.9674,
|
| 1927 |
+
"step": 27400
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 0.5732271647142202,
|
| 1931 |
+
"grad_norm": 1.1072654724121094,
|
| 1932 |
+
"learning_rate": 1.4330753797649757e-05,
|
| 1933 |
+
"loss": 1.9591,
|
| 1934 |
+
"step": 27500
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.5753116271313629,
|
| 1938 |
+
"grad_norm": 1.6574842929840088,
|
| 1939 |
+
"learning_rate": 1.4382865629641211e-05,
|
| 1940 |
+
"loss": 1.9743,
|
| 1941 |
+
"step": 27600
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.5773960895485054,
|
| 1945 |
+
"grad_norm": 1.6935937404632568,
|
| 1946 |
+
"learning_rate": 1.4434977461632665e-05,
|
| 1947 |
+
"loss": 1.9584,
|
| 1948 |
+
"step": 27700
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 0.5794805519656481,
|
| 1952 |
+
"grad_norm": 0.9915087819099426,
|
| 1953 |
+
"learning_rate": 1.4487089293624119e-05,
|
| 1954 |
+
"loss": 1.9661,
|
| 1955 |
+
"step": 27800
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 0.5815650143827906,
|
| 1959 |
+
"grad_norm": 1.6524442434310913,
|
| 1960 |
+
"learning_rate": 1.4539201125615571e-05,
|
| 1961 |
+
"loss": 1.9888,
|
| 1962 |
+
"step": 27900
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.5836494767999333,
|
| 1966 |
+
"grad_norm": 0.9779165983200073,
|
| 1967 |
+
"learning_rate": 1.4591312957607027e-05,
|
| 1968 |
+
"loss": 1.9528,
|
| 1969 |
+
"step": 28000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 0.585733939217076,
|
| 1973 |
+
"grad_norm": 1.5924351215362549,
|
| 1974 |
+
"learning_rate": 1.4643424789598479e-05,
|
| 1975 |
+
"loss": 1.9434,
|
| 1976 |
+
"step": 28100
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.5878184016342185,
|
| 1980 |
+
"grad_norm": 1.073769211769104,
|
| 1981 |
+
"learning_rate": 1.4695536621589935e-05,
|
| 1982 |
+
"loss": 1.9859,
|
| 1983 |
+
"step": 28200
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 0.5899028640513612,
|
| 1987 |
+
"grad_norm": 1.2847269773483276,
|
| 1988 |
+
"learning_rate": 1.4747648453581387e-05,
|
| 1989 |
+
"loss": 1.9691,
|
| 1990 |
+
"step": 28300
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 0.5919873264685038,
|
| 1994 |
+
"grad_norm": 2.1441261768341064,
|
| 1995 |
+
"learning_rate": 1.4799760285572839e-05,
|
| 1996 |
+
"loss": 1.9548,
|
| 1997 |
+
"step": 28400
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 0.5940717888856464,
|
| 2001 |
+
"grad_norm": 1.0148779153823853,
|
| 2002 |
+
"learning_rate": 1.4851872117564295e-05,
|
| 2003 |
+
"loss": 1.956,
|
| 2004 |
+
"step": 28500
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 0.596156251302789,
|
| 2008 |
+
"grad_norm": 1.3669894933700562,
|
| 2009 |
+
"learning_rate": 1.4903983949555747e-05,
|
| 2010 |
+
"loss": 1.9456,
|
| 2011 |
+
"step": 28600
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 0.5982407137199316,
|
| 2015 |
+
"grad_norm": 2.084087610244751,
|
| 2016 |
+
"learning_rate": 1.4956095781547202e-05,
|
| 2017 |
+
"loss": 1.9768,
|
| 2018 |
+
"step": 28700
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.6003251761370743,
|
| 2022 |
+
"grad_norm": 1.0451395511627197,
|
| 2023 |
+
"learning_rate": 1.5008207613538655e-05,
|
| 2024 |
+
"loss": 1.9628,
|
| 2025 |
+
"step": 28800
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 0.6024096385542169,
|
| 2029 |
+
"grad_norm": 1.0861454010009766,
|
| 2030 |
+
"learning_rate": 1.5060319445530109e-05,
|
| 2031 |
+
"loss": 1.9608,
|
| 2032 |
+
"step": 28900
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 0.6044941009713595,
|
| 2036 |
+
"grad_norm": 1.082926630973816,
|
| 2037 |
+
"learning_rate": 1.5112431277521562e-05,
|
| 2038 |
+
"loss": 1.9507,
|
| 2039 |
+
"step": 29000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 0.6065785633885021,
|
| 2043 |
+
"grad_norm": 1.0375463962554932,
|
| 2044 |
+
"learning_rate": 1.5164543109513016e-05,
|
| 2045 |
+
"loss": 1.9444,
|
| 2046 |
+
"step": 29100
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 0.6086630258056447,
|
| 2050 |
+
"grad_norm": 1.0970326662063599,
|
| 2051 |
+
"learning_rate": 1.521665494150447e-05,
|
| 2052 |
+
"loss": 1.9602,
|
| 2053 |
+
"step": 29200
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 0.6107474882227873,
|
| 2057 |
+
"grad_norm": 1.4305795431137085,
|
| 2058 |
+
"learning_rate": 1.5268766773495924e-05,
|
| 2059 |
+
"loss": 1.9663,
|
| 2060 |
+
"step": 29300
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 0.61283195063993,
|
| 2064 |
+
"grad_norm": 1.2782924175262451,
|
| 2065 |
+
"learning_rate": 1.5320878605487378e-05,
|
| 2066 |
+
"loss": 1.9706,
|
| 2067 |
+
"step": 29400
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 0.6149164130570726,
|
| 2071 |
+
"grad_norm": 1.2696731090545654,
|
| 2072 |
+
"learning_rate": 1.5372990437478832e-05,
|
| 2073 |
+
"loss": 1.9502,
|
| 2074 |
+
"step": 29500
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.6170008754742152,
|
| 2078 |
+
"grad_norm": 1.0566843748092651,
|
| 2079 |
+
"learning_rate": 1.5425102269470286e-05,
|
| 2080 |
+
"loss": 1.948,
|
| 2081 |
+
"step": 29600
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 0.6190853378913578,
|
| 2085 |
+
"grad_norm": 1.1071428060531616,
|
| 2086 |
+
"learning_rate": 1.5477214101461736e-05,
|
| 2087 |
+
"loss": 1.9409,
|
| 2088 |
+
"step": 29700
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 0.6211698003085004,
|
| 2092 |
+
"grad_norm": 0.9255651235580444,
|
| 2093 |
+
"learning_rate": 1.5529325933453194e-05,
|
| 2094 |
+
"loss": 1.9466,
|
| 2095 |
+
"step": 29800
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 0.6232542627256431,
|
| 2099 |
+
"grad_norm": 1.0186619758605957,
|
| 2100 |
+
"learning_rate": 1.5581437765444644e-05,
|
| 2101 |
+
"loss": 1.9493,
|
| 2102 |
+
"step": 29900
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.6253387251427857,
|
| 2106 |
+
"grad_norm": 1.451917052268982,
|
| 2107 |
+
"learning_rate": 1.5633549597436098e-05,
|
| 2108 |
+
"loss": 1.9503,
|
| 2109 |
+
"step": 30000
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 0.6274231875599283,
|
| 2113 |
+
"grad_norm": 1.0187926292419434,
|
| 2114 |
+
"learning_rate": 1.5685661429427552e-05,
|
| 2115 |
+
"loss": 1.9563,
|
| 2116 |
+
"step": 30100
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.6295076499770709,
|
| 2120 |
+
"grad_norm": 0.871369481086731,
|
| 2121 |
+
"learning_rate": 1.5737773261419006e-05,
|
| 2122 |
+
"loss": 1.9549,
|
| 2123 |
+
"step": 30200
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 0.6315921123942135,
|
| 2127 |
+
"grad_norm": 0.98649662733078,
|
| 2128 |
+
"learning_rate": 1.578988509341046e-05,
|
| 2129 |
+
"loss": 1.9596,
|
| 2130 |
+
"step": 30300
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 0.6336765748113562,
|
| 2134 |
+
"grad_norm": 1.6009703874588013,
|
| 2135 |
+
"learning_rate": 1.5841996925401914e-05,
|
| 2136 |
+
"loss": 1.9348,
|
| 2137 |
+
"step": 30400
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 0.6357610372284987,
|
| 2141 |
+
"grad_norm": 1.0804214477539062,
|
| 2142 |
+
"learning_rate": 1.5894108757393368e-05,
|
| 2143 |
+
"loss": 1.9428,
|
| 2144 |
+
"step": 30500
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 0.6378454996456414,
|
| 2148 |
+
"grad_norm": 1.1061315536499023,
|
| 2149 |
+
"learning_rate": 1.594622058938482e-05,
|
| 2150 |
+
"loss": 1.9371,
|
| 2151 |
+
"step": 30600
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 0.639929962062784,
|
| 2155 |
+
"grad_norm": 1.0033613443374634,
|
| 2156 |
+
"learning_rate": 1.5998332421376275e-05,
|
| 2157 |
+
"loss": 1.9508,
|
| 2158 |
+
"step": 30700
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 0.6420144244799266,
|
| 2162 |
+
"grad_norm": 0.9282180070877075,
|
| 2163 |
+
"learning_rate": 1.6050444253367726e-05,
|
| 2164 |
+
"loss": 1.9376,
|
| 2165 |
+
"step": 30800
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 0.6440988868970693,
|
| 2169 |
+
"grad_norm": 0.9301239252090454,
|
| 2170 |
+
"learning_rate": 1.6102556085359183e-05,
|
| 2171 |
+
"loss": 1.9556,
|
| 2172 |
+
"step": 30900
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 0.6461833493142118,
|
| 2176 |
+
"grad_norm": 0.9229454398155212,
|
| 2177 |
+
"learning_rate": 1.6154667917350634e-05,
|
| 2178 |
+
"loss": 1.9314,
|
| 2179 |
+
"step": 31000
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 0.6482678117313545,
|
| 2183 |
+
"grad_norm": 1.0757591724395752,
|
| 2184 |
+
"learning_rate": 1.620677974934209e-05,
|
| 2185 |
+
"loss": 1.9309,
|
| 2186 |
+
"step": 31100
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 0.6503522741484971,
|
| 2190 |
+
"grad_norm": 3.362175464630127,
|
| 2191 |
+
"learning_rate": 1.625889158133354e-05,
|
| 2192 |
+
"loss": 1.9254,
|
| 2193 |
+
"step": 31200
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 0.6524367365656397,
|
| 2197 |
+
"grad_norm": 1.1901942491531372,
|
| 2198 |
+
"learning_rate": 1.6311003413325e-05,
|
| 2199 |
+
"loss": 1.9337,
|
| 2200 |
+
"step": 31300
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 0.6545211989827824,
|
| 2204 |
+
"grad_norm": 1.2026360034942627,
|
| 2205 |
+
"learning_rate": 1.636311524531645e-05,
|
| 2206 |
+
"loss": 1.9407,
|
| 2207 |
+
"step": 31400
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 0.6566056613999249,
|
| 2211 |
+
"grad_norm": 1.247865915298462,
|
| 2212 |
+
"learning_rate": 1.6415227077307903e-05,
|
| 2213 |
+
"loss": 1.9345,
|
| 2214 |
+
"step": 31500
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 0.6586901238170676,
|
| 2218 |
+
"grad_norm": 1.0052815675735474,
|
| 2219 |
+
"learning_rate": 1.6467338909299357e-05,
|
| 2220 |
+
"loss": 1.9295,
|
| 2221 |
+
"step": 31600
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 0.6607745862342101,
|
| 2225 |
+
"grad_norm": 0.9365499019622803,
|
| 2226 |
+
"learning_rate": 1.651945074129081e-05,
|
| 2227 |
+
"loss": 1.9421,
|
| 2228 |
+
"step": 31700
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 0.6628590486513528,
|
| 2232 |
+
"grad_norm": 1.7018957138061523,
|
| 2233 |
+
"learning_rate": 1.6571562573282265e-05,
|
| 2234 |
+
"loss": 1.9174,
|
| 2235 |
+
"step": 31800
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 0.6649435110684955,
|
| 2239 |
+
"grad_norm": 1.0856178998947144,
|
| 2240 |
+
"learning_rate": 1.662367440527372e-05,
|
| 2241 |
+
"loss": 1.9045,
|
| 2242 |
+
"step": 31900
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.667027973485638,
|
| 2246 |
+
"grad_norm": 1.111540675163269,
|
| 2247 |
+
"learning_rate": 1.6675786237265173e-05,
|
| 2248 |
+
"loss": 1.9263,
|
| 2249 |
+
"step": 32000
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 0.6691124359027807,
|
| 2253 |
+
"grad_norm": 0.9781438708305359,
|
| 2254 |
+
"learning_rate": 1.6727898069256626e-05,
|
| 2255 |
+
"loss": 1.9115,
|
| 2256 |
+
"step": 32100
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 0.6711968983199232,
|
| 2260 |
+
"grad_norm": 0.9509501457214355,
|
| 2261 |
+
"learning_rate": 1.678000990124808e-05,
|
| 2262 |
+
"loss": 1.9194,
|
| 2263 |
+
"step": 32200
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 0.6732813607370659,
|
| 2267 |
+
"grad_norm": 1.0077457427978516,
|
| 2268 |
+
"learning_rate": 1.6832121733239534e-05,
|
| 2269 |
+
"loss": 1.9361,
|
| 2270 |
+
"step": 32300
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 0.6753658231542086,
|
| 2274 |
+
"grad_norm": 1.0101159811019897,
|
| 2275 |
+
"learning_rate": 1.6884233565230988e-05,
|
| 2276 |
+
"loss": 1.9227,
|
| 2277 |
+
"step": 32400
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 0.6774502855713511,
|
| 2281 |
+
"grad_norm": 1.6001538038253784,
|
| 2282 |
+
"learning_rate": 1.693634539722244e-05,
|
| 2283 |
+
"loss": 1.9114,
|
| 2284 |
+
"step": 32500
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 0.6795347479884938,
|
| 2288 |
+
"grad_norm": 1.300024151802063,
|
| 2289 |
+
"learning_rate": 1.6988457229213896e-05,
|
| 2290 |
+
"loss": 1.9471,
|
| 2291 |
+
"step": 32600
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 0.6816192104056363,
|
| 2295 |
+
"grad_norm": 1.0062751770019531,
|
| 2296 |
+
"learning_rate": 1.7040569061205346e-05,
|
| 2297 |
+
"loss": 1.9043,
|
| 2298 |
+
"step": 32700
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 0.683703672822779,
|
| 2302 |
+
"grad_norm": 1.012172818183899,
|
| 2303 |
+
"learning_rate": 1.7092680893196804e-05,
|
| 2304 |
+
"loss": 1.9189,
|
| 2305 |
+
"step": 32800
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 0.6857881352399217,
|
| 2309 |
+
"grad_norm": 2.5829713344573975,
|
| 2310 |
+
"learning_rate": 1.7144792725188254e-05,
|
| 2311 |
+
"loss": 1.8859,
|
| 2312 |
+
"step": 32900
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 0.6878725976570642,
|
| 2316 |
+
"grad_norm": 1.1416168212890625,
|
| 2317 |
+
"learning_rate": 1.719690455717971e-05,
|
| 2318 |
+
"loss": 1.9426,
|
| 2319 |
+
"step": 33000
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 0.6899570600742069,
|
| 2323 |
+
"grad_norm": 5.545751094818115,
|
| 2324 |
+
"learning_rate": 1.7249016389171162e-05,
|
| 2325 |
+
"loss": 1.9544,
|
| 2326 |
+
"step": 33100
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 0.6920415224913494,
|
| 2330 |
+
"grad_norm": 1.051398515701294,
|
| 2331 |
+
"learning_rate": 1.7301128221162616e-05,
|
| 2332 |
+
"loss": 1.9391,
|
| 2333 |
+
"step": 33200
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 0.6941259849084921,
|
| 2337 |
+
"grad_norm": 1.153196930885315,
|
| 2338 |
+
"learning_rate": 1.735324005315407e-05,
|
| 2339 |
+
"loss": 1.9112,
|
| 2340 |
+
"step": 33300
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 0.6962104473256348,
|
| 2344 |
+
"grad_norm": 1.1652779579162598,
|
| 2345 |
+
"learning_rate": 1.7405351885145524e-05,
|
| 2346 |
+
"loss": 1.9035,
|
| 2347 |
+
"step": 33400
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 0.6982949097427773,
|
| 2351 |
+
"grad_norm": 1.9276180267333984,
|
| 2352 |
+
"learning_rate": 1.7457463717136978e-05,
|
| 2353 |
+
"loss": 1.9415,
|
| 2354 |
+
"step": 33500
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 0.70037937215992,
|
| 2358 |
+
"grad_norm": 0.8301405310630798,
|
| 2359 |
+
"learning_rate": 1.750957554912843e-05,
|
| 2360 |
+
"loss": 1.9054,
|
| 2361 |
+
"step": 33600
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 0.7024638345770626,
|
| 2365 |
+
"grad_norm": 0.9777198433876038,
|
| 2366 |
+
"learning_rate": 1.7561687381119885e-05,
|
| 2367 |
+
"loss": 1.8924,
|
| 2368 |
+
"step": 33700
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 0.7045482969942052,
|
| 2372 |
+
"grad_norm": 1.2608678340911865,
|
| 2373 |
+
"learning_rate": 1.761379921311134e-05,
|
| 2374 |
+
"loss": 1.8982,
|
| 2375 |
+
"step": 33800
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 0.7066327594113478,
|
| 2379 |
+
"grad_norm": 0.9381488561630249,
|
| 2380 |
+
"learning_rate": 1.7665911045102793e-05,
|
| 2381 |
+
"loss": 1.89,
|
| 2382 |
+
"step": 33900
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 0.7087172218284904,
|
| 2386 |
+
"grad_norm": 1.045607328414917,
|
| 2387 |
+
"learning_rate": 1.7718022877094244e-05,
|
| 2388 |
+
"loss": 1.9073,
|
| 2389 |
+
"step": 34000
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 0.7108016842456331,
|
| 2393 |
+
"grad_norm": 1.0266004800796509,
|
| 2394 |
+
"learning_rate": 1.77701347090857e-05,
|
| 2395 |
+
"loss": 1.9222,
|
| 2396 |
+
"step": 34100
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 0.7128861466627757,
|
| 2400 |
+
"grad_norm": 1.1733341217041016,
|
| 2401 |
+
"learning_rate": 1.782224654107715e-05,
|
| 2402 |
+
"loss": 1.9041,
|
| 2403 |
+
"step": 34200
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 0.7149706090799183,
|
| 2407 |
+
"grad_norm": 1.0179702043533325,
|
| 2408 |
+
"learning_rate": 1.787435837306861e-05,
|
| 2409 |
+
"loss": 1.9133,
|
| 2410 |
+
"step": 34300
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 0.7170550714970609,
|
| 2414 |
+
"grad_norm": 1.2114508152008057,
|
| 2415 |
+
"learning_rate": 1.792647020506006e-05,
|
| 2416 |
+
"loss": 1.8858,
|
| 2417 |
+
"step": 34400
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 0.7191395339142035,
|
| 2421 |
+
"grad_norm": 0.9730684161186218,
|
| 2422 |
+
"learning_rate": 1.7978582037051513e-05,
|
| 2423 |
+
"loss": 1.9143,
|
| 2424 |
+
"step": 34500
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 0.7212239963313462,
|
| 2428 |
+
"grad_norm": 0.9815762639045715,
|
| 2429 |
+
"learning_rate": 1.8030693869042967e-05,
|
| 2430 |
+
"loss": 1.9032,
|
| 2431 |
+
"step": 34600
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 0.7233084587484888,
|
| 2435 |
+
"grad_norm": 0.979191243648529,
|
| 2436 |
+
"learning_rate": 1.808280570103442e-05,
|
| 2437 |
+
"loss": 1.8858,
|
| 2438 |
+
"step": 34700
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 0.7253929211656314,
|
| 2442 |
+
"grad_norm": 1.051487684249878,
|
| 2443 |
+
"learning_rate": 1.8134917533025875e-05,
|
| 2444 |
+
"loss": 1.8935,
|
| 2445 |
+
"step": 34800
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 0.727477383582774,
|
| 2449 |
+
"grad_norm": 1.1583081483840942,
|
| 2450 |
+
"learning_rate": 1.818702936501733e-05,
|
| 2451 |
+
"loss": 1.8998,
|
| 2452 |
+
"step": 34900
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 0.7295618459999166,
|
| 2456 |
+
"grad_norm": 1.3905084133148193,
|
| 2457 |
+
"learning_rate": 1.8239141197008783e-05,
|
| 2458 |
+
"loss": 1.9105,
|
| 2459 |
+
"step": 35000
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 0.7316463084170592,
|
| 2463 |
+
"grad_norm": 0.9211100339889526,
|
| 2464 |
+
"learning_rate": 1.8291253029000237e-05,
|
| 2465 |
+
"loss": 1.9145,
|
| 2466 |
+
"step": 35100
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 0.7337307708342019,
|
| 2470 |
+
"grad_norm": 0.9945178627967834,
|
| 2471 |
+
"learning_rate": 1.834336486099169e-05,
|
| 2472 |
+
"loss": 1.9049,
|
| 2473 |
+
"step": 35200
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 0.7358152332513445,
|
| 2477 |
+
"grad_norm": 0.9826840162277222,
|
| 2478 |
+
"learning_rate": 1.8395476692983144e-05,
|
| 2479 |
+
"loss": 1.8968,
|
| 2480 |
+
"step": 35300
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 0.7378996956684871,
|
| 2484 |
+
"grad_norm": 1.0781490802764893,
|
| 2485 |
+
"learning_rate": 1.8447588524974598e-05,
|
| 2486 |
+
"loss": 1.9173,
|
| 2487 |
+
"step": 35400
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 0.7399841580856297,
|
| 2491 |
+
"grad_norm": 1.347640037536621,
|
| 2492 |
+
"learning_rate": 1.849970035696605e-05,
|
| 2493 |
+
"loss": 1.8976,
|
| 2494 |
+
"step": 35500
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.7420686205027723,
|
| 2498 |
+
"grad_norm": 0.9135119318962097,
|
| 2499 |
+
"learning_rate": 1.8551812188957506e-05,
|
| 2500 |
+
"loss": 1.8965,
|
| 2501 |
+
"step": 35600
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 0.744153082919915,
|
| 2505 |
+
"grad_norm": 1.1346653699874878,
|
| 2506 |
+
"learning_rate": 1.8603924020948957e-05,
|
| 2507 |
+
"loss": 1.8989,
|
| 2508 |
+
"step": 35700
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 0.7462375453370576,
|
| 2512 |
+
"grad_norm": 1.0951207876205444,
|
| 2513 |
+
"learning_rate": 1.865603585294041e-05,
|
| 2514 |
+
"loss": 1.8982,
|
| 2515 |
+
"step": 35800
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 0.7483220077542002,
|
| 2519 |
+
"grad_norm": 1.0009814500808716,
|
| 2520 |
+
"learning_rate": 1.8708147684931864e-05,
|
| 2521 |
+
"loss": 1.8914,
|
| 2522 |
+
"step": 35900
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 0.7504064701713428,
|
| 2526 |
+
"grad_norm": 0.9622611403465271,
|
| 2527 |
+
"learning_rate": 1.8760259516923318e-05,
|
| 2528 |
+
"loss": 1.8844,
|
| 2529 |
+
"step": 36000
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 0.7524909325884854,
|
| 2533 |
+
"grad_norm": 1.4782415628433228,
|
| 2534 |
+
"learning_rate": 1.8812371348914772e-05,
|
| 2535 |
+
"loss": 1.8879,
|
| 2536 |
+
"step": 36100
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 0.7545753950056281,
|
| 2540 |
+
"grad_norm": 1.0756487846374512,
|
| 2541 |
+
"learning_rate": 1.8864483180906226e-05,
|
| 2542 |
+
"loss": 1.908,
|
| 2543 |
+
"step": 36200
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 0.7566598574227706,
|
| 2547 |
+
"grad_norm": 1.1616325378417969,
|
| 2548 |
+
"learning_rate": 1.891659501289768e-05,
|
| 2549 |
+
"loss": 1.9068,
|
| 2550 |
+
"step": 36300
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 0.7587443198399133,
|
| 2554 |
+
"grad_norm": 1.1253652572631836,
|
| 2555 |
+
"learning_rate": 1.8968706844889134e-05,
|
| 2556 |
+
"loss": 1.9044,
|
| 2557 |
+
"step": 36400
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 0.7608287822570559,
|
| 2561 |
+
"grad_norm": 0.9540199041366577,
|
| 2562 |
+
"learning_rate": 1.9020818676880588e-05,
|
| 2563 |
+
"loss": 1.857,
|
| 2564 |
+
"step": 36500
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.7629132446741985,
|
| 2568 |
+
"grad_norm": 1.3189705610275269,
|
| 2569 |
+
"learning_rate": 1.907293050887204e-05,
|
| 2570 |
+
"loss": 1.8842,
|
| 2571 |
+
"step": 36600
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 0.7649977070913412,
|
| 2575 |
+
"grad_norm": 1.7837036848068237,
|
| 2576 |
+
"learning_rate": 1.9125042340863496e-05,
|
| 2577 |
+
"loss": 1.861,
|
| 2578 |
+
"step": 36700
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 0.7670821695084837,
|
| 2582 |
+
"grad_norm": 1.01956307888031,
|
| 2583 |
+
"learning_rate": 1.917715417285495e-05,
|
| 2584 |
+
"loss": 1.8852,
|
| 2585 |
+
"step": 36800
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 0.7691666319256264,
|
| 2589 |
+
"grad_norm": 1.1640490293502808,
|
| 2590 |
+
"learning_rate": 1.9229266004846403e-05,
|
| 2591 |
+
"loss": 1.8695,
|
| 2592 |
+
"step": 36900
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 0.771251094342769,
|
| 2596 |
+
"grad_norm": 1.3748284578323364,
|
| 2597 |
+
"learning_rate": 1.9281377836837857e-05,
|
| 2598 |
+
"loss": 1.893,
|
| 2599 |
+
"step": 37000
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 0.7733355567599116,
|
| 2603 |
+
"grad_norm": 0.8749508261680603,
|
| 2604 |
+
"learning_rate": 1.9333489668829308e-05,
|
| 2605 |
+
"loss": 1.8791,
|
| 2606 |
+
"step": 37100
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 0.7754200191770543,
|
| 2610 |
+
"grad_norm": 0.9299560785293579,
|
| 2611 |
+
"learning_rate": 1.938560150082076e-05,
|
| 2612 |
+
"loss": 1.8808,
|
| 2613 |
+
"step": 37200
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 0.7775044815941968,
|
| 2617 |
+
"grad_norm": 0.8478776216506958,
|
| 2618 |
+
"learning_rate": 1.9437713332812216e-05,
|
| 2619 |
+
"loss": 1.8827,
|
| 2620 |
+
"step": 37300
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 0.7795889440113395,
|
| 2624 |
+
"grad_norm": 1.0107454061508179,
|
| 2625 |
+
"learning_rate": 1.948982516480367e-05,
|
| 2626 |
+
"loss": 1.9092,
|
| 2627 |
+
"step": 37400
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 0.781673406428482,
|
| 2631 |
+
"grad_norm": 1.0087549686431885,
|
| 2632 |
+
"learning_rate": 1.9541936996795123e-05,
|
| 2633 |
+
"loss": 1.8792,
|
| 2634 |
+
"step": 37500
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 0.7837578688456247,
|
| 2638 |
+
"grad_norm": 1.1416882276535034,
|
| 2639 |
+
"learning_rate": 1.9594048828786577e-05,
|
| 2640 |
+
"loss": 1.8796,
|
| 2641 |
+
"step": 37600
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 0.7858423312627674,
|
| 2645 |
+
"grad_norm": 0.9162054657936096,
|
| 2646 |
+
"learning_rate": 1.964616066077803e-05,
|
| 2647 |
+
"loss": 1.8829,
|
| 2648 |
+
"step": 37700
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 0.7879267936799099,
|
| 2652 |
+
"grad_norm": 0.990573525428772,
|
| 2653 |
+
"learning_rate": 1.9698272492769485e-05,
|
| 2654 |
+
"loss": 1.8859,
|
| 2655 |
+
"step": 37800
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 0.7900112560970526,
|
| 2659 |
+
"grad_norm": 0.8790944814682007,
|
| 2660 |
+
"learning_rate": 1.975038432476094e-05,
|
| 2661 |
+
"loss": 1.8789,
|
| 2662 |
+
"step": 37900
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 0.7920957185141951,
|
| 2666 |
+
"grad_norm": 2.4909720420837402,
|
| 2667 |
+
"learning_rate": 1.9802496156752393e-05,
|
| 2668 |
+
"loss": 1.9049,
|
| 2669 |
+
"step": 38000
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 0.7941801809313378,
|
| 2673 |
+
"grad_norm": 1.4410492181777954,
|
| 2674 |
+
"learning_rate": 1.9854607988743847e-05,
|
| 2675 |
+
"loss": 1.8903,
|
| 2676 |
+
"step": 38100
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 0.7962646433484805,
|
| 2680 |
+
"grad_norm": 1.854480266571045,
|
| 2681 |
+
"learning_rate": 1.9906719820735297e-05,
|
| 2682 |
+
"loss": 1.8948,
|
| 2683 |
+
"step": 38200
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 0.798349105765623,
|
| 2687 |
+
"grad_norm": 0.9377720952033997,
|
| 2688 |
+
"learning_rate": 1.9958831652726755e-05,
|
| 2689 |
+
"loss": 1.8835,
|
| 2690 |
+
"step": 38300
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 0.8004335681827657,
|
| 2694 |
+
"grad_norm": 0.863426148891449,
|
| 2695 |
+
"learning_rate": 1.9999763616765983e-05,
|
| 2696 |
+
"loss": 1.888,
|
| 2697 |
+
"step": 38400
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 0.8025180305999083,
|
| 2701 |
+
"grad_norm": 0.884555459022522,
|
| 2702 |
+
"learning_rate": 1.9992153179340215e-05,
|
| 2703 |
+
"loss": 1.8643,
|
| 2704 |
+
"step": 38500
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 0.8046024930170509,
|
| 2708 |
+
"grad_norm": 1.795231819152832,
|
| 2709 |
+
"learning_rate": 1.9973831740621872e-05,
|
| 2710 |
+
"loss": 1.8671,
|
| 2711 |
+
"step": 38600
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 0.8066869554341936,
|
| 2715 |
+
"grad_norm": 0.8534547090530396,
|
| 2716 |
+
"learning_rate": 1.9944818940117093e-05,
|
| 2717 |
+
"loss": 1.8869,
|
| 2718 |
+
"step": 38700
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 0.8087714178513361,
|
| 2722 |
+
"grad_norm": 0.8905624747276306,
|
| 2723 |
+
"learning_rate": 1.9905145877843883e-05,
|
| 2724 |
+
"loss": 1.8813,
|
| 2725 |
+
"step": 38800
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 0.8108558802684788,
|
| 2729 |
+
"grad_norm": 1.0225939750671387,
|
| 2730 |
+
"learning_rate": 1.9854855080994695e-05,
|
| 2731 |
+
"loss": 1.8797,
|
| 2732 |
+
"step": 38900
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 0.8129403426856214,
|
| 2736 |
+
"grad_norm": 1.0849330425262451,
|
| 2737 |
+
"learning_rate": 1.9794000458349786e-05,
|
| 2738 |
+
"loss": 1.8685,
|
| 2739 |
+
"step": 39000
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 0.815024805102764,
|
| 2743 |
+
"grad_norm": 1.1683331727981567,
|
| 2744 |
+
"learning_rate": 1.9722647242490173e-05,
|
| 2745 |
+
"loss": 1.8658,
|
| 2746 |
+
"step": 39100
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 0.8171092675199066,
|
| 2750 |
+
"grad_norm": 2.3962318897247314,
|
| 2751 |
+
"learning_rate": 1.9640871919872137e-05,
|
| 2752 |
+
"loss": 1.8733,
|
| 2753 |
+
"step": 39200
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 0.8191937299370492,
|
| 2757 |
+
"grad_norm": 0.8535931706428528,
|
| 2758 |
+
"learning_rate": 1.9548762148838235e-05,
|
| 2759 |
+
"loss": 1.8777,
|
| 2760 |
+
"step": 39300
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 0.8212781923541919,
|
| 2764 |
+
"grad_norm": 1.10763418674469,
|
| 2765 |
+
"learning_rate": 1.94464166656527e-05,
|
| 2766 |
+
"loss": 1.8821,
|
| 2767 |
+
"step": 39400
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 0.8233626547713345,
|
| 2771 |
+
"grad_norm": 5.269350528717041,
|
| 2772 |
+
"learning_rate": 1.933394517866198e-05,
|
| 2773 |
+
"loss": 1.8522,
|
| 2774 |
+
"step": 39500
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 0.8254471171884771,
|
| 2778 |
+
"grad_norm": 1.2906358242034912,
|
| 2779 |
+
"learning_rate": 1.9211468250693814e-05,
|
| 2780 |
+
"loss": 1.8851,
|
| 2781 |
+
"step": 39600
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 0.8275315796056197,
|
| 2785 |
+
"grad_norm": 1.466866374015808,
|
| 2786 |
+
"learning_rate": 1.907911716982097e-05,
|
| 2787 |
+
"loss": 1.8516,
|
| 2788 |
+
"step": 39700
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 0.8296160420227623,
|
| 2792 |
+
"grad_norm": 0.8515104651451111,
|
| 2793 |
+
"learning_rate": 1.8937033808628157e-05,
|
| 2794 |
+
"loss": 1.8722,
|
| 2795 |
+
"step": 39800
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 0.831700504439905,
|
| 2799 |
+
"grad_norm": 0.8316900134086609,
|
| 2800 |
+
"learning_rate": 1.8785370472132925e-05,
|
| 2801 |
+
"loss": 1.8794,
|
| 2802 |
+
"step": 39900
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 0.8337849668570476,
|
| 2806 |
+
"grad_norm": 2.936185598373413,
|
| 2807 |
+
"learning_rate": 1.8624289734523655e-05,
|
| 2808 |
+
"loss": 1.8766,
|
| 2809 |
+
"step": 40000
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 0.8358694292741902,
|
| 2813 |
+
"grad_norm": 0.8948720693588257,
|
| 2814 |
+
"learning_rate": 1.8453964264889555e-05,
|
| 2815 |
+
"loss": 1.8679,
|
| 2816 |
+
"step": 40100
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 0.8379538916913328,
|
| 2820 |
+
"grad_norm": 1.0251227617263794,
|
| 2821 |
+
"learning_rate": 1.8274576642129552e-05,
|
| 2822 |
+
"loss": 1.8599,
|
| 2823 |
+
"step": 40200
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 0.8400383541084754,
|
| 2827 |
+
"grad_norm": 2.930690050125122,
|
| 2828 |
+
"learning_rate": 1.8086319159238442e-05,
|
| 2829 |
+
"loss": 1.8634,
|
| 2830 |
+
"step": 40300
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 0.842122816525618,
|
| 2834 |
+
"grad_norm": 1.140226125717163,
|
| 2835 |
+
"learning_rate": 1.7889393617180074e-05,
|
| 2836 |
+
"loss": 1.8637,
|
| 2837 |
+
"step": 40400
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 0.8442072789427607,
|
| 2841 |
+
"grad_norm": 0.9171246290206909,
|
| 2842 |
+
"learning_rate": 1.7684011108568593e-05,
|
| 2843 |
+
"loss": 1.854,
|
| 2844 |
+
"step": 40500
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 0.8462917413599033,
|
| 2848 |
+
"grad_norm": 0.9319982528686523,
|
| 2849 |
+
"learning_rate": 1.747039179138954e-05,
|
| 2850 |
+
"loss": 1.8384,
|
| 2851 |
+
"step": 40600
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 0.8483762037770459,
|
| 2855 |
+
"grad_norm": 1.0997920036315918,
|
| 2856 |
+
"learning_rate": 1.7248764653003433e-05,
|
| 2857 |
+
"loss": 1.8613,
|
| 2858 |
+
"step": 40700
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 0.8504606661941885,
|
| 2862 |
+
"grad_norm": 0.8980279564857483,
|
| 2863 |
+
"learning_rate": 1.701936726468477e-05,
|
| 2864 |
+
"loss": 1.8864,
|
| 2865 |
+
"step": 40800
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 0.8525451286113311,
|
| 2869 |
+
"grad_norm": 1.1140270233154297,
|
| 2870 |
+
"learning_rate": 1.6782445526959553e-05,
|
| 2871 |
+
"loss": 1.8564,
|
| 2872 |
+
"step": 40900
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 0.8546295910284738,
|
| 2876 |
+
"grad_norm": 0.9438982009887695,
|
| 2877 |
+
"learning_rate": 1.65382534060144e-05,
|
| 2878 |
+
"loss": 1.8686,
|
| 2879 |
+
"step": 41000
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 0.8567140534456164,
|
| 2883 |
+
"grad_norm": 1.087593913078308,
|
| 2884 |
+
"learning_rate": 1.628705266145969e-05,
|
| 2885 |
+
"loss": 1.8612,
|
| 2886 |
+
"step": 41100
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 0.858798515862759,
|
| 2890 |
+
"grad_norm": 1.5680657625198364,
|
| 2891 |
+
"learning_rate": 1.6029112565738647e-05,
|
| 2892 |
+
"loss": 1.8595,
|
| 2893 |
+
"step": 41200
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 0.8608829782799016,
|
| 2897 |
+
"grad_norm": 1.1920835971832275,
|
| 2898 |
+
"learning_rate": 1.5764709615483084e-05,
|
| 2899 |
+
"loss": 1.8368,
|
| 2900 |
+
"step": 41300
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 0.8629674406970442,
|
| 2904 |
+
"grad_norm": 0.9491363167762756,
|
| 2905 |
+
"learning_rate": 1.549412723512526e-05,
|
| 2906 |
+
"loss": 1.8433,
|
| 2907 |
+
"step": 41400
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 0.8650519031141869,
|
| 2911 |
+
"grad_norm": 0.8679408431053162,
|
| 2912 |
+
"learning_rate": 1.521765547308355e-05,
|
| 2913 |
+
"loss": 1.8585,
|
| 2914 |
+
"step": 41500
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 0.8671363655313294,
|
| 2918 |
+
"grad_norm": 0.9135258793830872,
|
| 2919 |
+
"learning_rate": 1.4935590690847555e-05,
|
| 2920 |
+
"loss": 1.8535,
|
| 2921 |
+
"step": 41600
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 0.8692208279484721,
|
| 2925 |
+
"grad_norm": 0.9444493055343628,
|
| 2926 |
+
"learning_rate": 1.464823524529601e-05,
|
| 2927 |
+
"loss": 1.8606,
|
| 2928 |
+
"step": 41700
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 0.8713052903656147,
|
| 2932 |
+
"grad_norm": 0.9098984003067017,
|
| 2933 |
+
"learning_rate": 1.4355897164587957e-05,
|
| 2934 |
+
"loss": 1.8634,
|
| 2935 |
+
"step": 41800
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 0.8733897527827573,
|
| 2939 |
+
"grad_norm": 4.627810478210449,
|
| 2940 |
+
"learning_rate": 1.4058889817974664e-05,
|
| 2941 |
+
"loss": 1.8349,
|
| 2942 |
+
"step": 41900
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 0.8754742151999,
|
| 2946 |
+
"grad_norm": 0.8299252986907959,
|
| 2947 |
+
"learning_rate": 1.3757531579886186e-05,
|
| 2948 |
+
"loss": 1.8541,
|
| 2949 |
+
"step": 42000
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 0.8775586776170425,
|
| 2953 |
+
"grad_norm": 0.9095094203948975,
|
| 2954 |
+
"learning_rate": 1.345214548865267e-05,
|
| 2955 |
+
"loss": 1.885,
|
| 2956 |
+
"step": 42100
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 0.8796431400341852,
|
| 2960 |
+
"grad_norm": 0.9644302725791931,
|
| 2961 |
+
"learning_rate": 1.3143058900226262e-05,
|
| 2962 |
+
"loss": 1.8788,
|
| 2963 |
+
"step": 42200
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 0.8817276024513278,
|
| 2967 |
+
"grad_norm": 1.4231350421905518,
|
| 2968 |
+
"learning_rate": 1.2830603137274729e-05,
|
| 2969 |
+
"loss": 1.8451,
|
| 2970 |
+
"step": 42300
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 0.8838120648684704,
|
| 2974 |
+
"grad_norm": 1.0535882711410522,
|
| 2975 |
+
"learning_rate": 1.2515113134023035e-05,
|
| 2976 |
+
"loss": 1.864,
|
| 2977 |
+
"step": 42400
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 0.8858965272856131,
|
| 2981 |
+
"grad_norm": 1.3644357919692993,
|
| 2982 |
+
"learning_rate": 1.2196927077223523e-05,
|
| 2983 |
+
"loss": 1.853,
|
| 2984 |
+
"step": 42500
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 0.8879809897027556,
|
| 2988 |
+
"grad_norm": 0.9575700163841248,
|
| 2989 |
+
"learning_rate": 1.187638604363958e-05,
|
| 2990 |
+
"loss": 1.851,
|
| 2991 |
+
"step": 42600
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 0.8900654521198983,
|
| 2995 |
+
"grad_norm": 1.0932910442352295,
|
| 2996 |
+
"learning_rate": 1.1553833634431375e-05,
|
| 2997 |
+
"loss": 1.8539,
|
| 2998 |
+
"step": 42700
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 0.8921499145370408,
|
| 3002 |
+
"grad_norm": 0.8120511174201965,
|
| 3003 |
+
"learning_rate": 1.1229615606835609e-05,
|
| 3004 |
+
"loss": 1.8331,
|
| 3005 |
+
"step": 42800
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 0.8942343769541835,
|
| 3009 |
+
"grad_norm": 0.967151939868927,
|
| 3010 |
+
"learning_rate": 1.0904079503534057e-05,
|
| 3011 |
+
"loss": 1.8565,
|
| 3012 |
+
"step": 42900
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 0.8963188393713262,
|
| 3016 |
+
"grad_norm": 0.8279860019683838,
|
| 3017 |
+
"learning_rate": 1.0577574280108226e-05,
|
| 3018 |
+
"loss": 1.8469,
|
| 3019 |
+
"step": 43000
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 0.8984033017884687,
|
| 3023 |
+
"grad_norm": 0.986449658870697,
|
| 3024 |
+
"learning_rate": 1.0250449930979484e-05,
|
| 3025 |
+
"loss": 1.8506,
|
| 3026 |
+
"step": 43100
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 0.9004877642056114,
|
| 3030 |
+
"grad_norm": 0.9188694953918457,
|
| 3031 |
+
"learning_rate": 9.92305711423557e-06,
|
| 3032 |
+
"loss": 1.8345,
|
| 3033 |
+
"step": 43200
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 0.9025722266227539,
|
| 3037 |
+
"grad_norm": 0.8622938990592957,
|
| 3038 |
+
"learning_rate": 9.595746775745753e-06,
|
| 3039 |
+
"loss": 1.8451,
|
| 3040 |
+
"step": 43300
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 0.9046566890398966,
|
| 3044 |
+
"grad_norm": 1.5604217052459717,
|
| 3045 |
+
"learning_rate": 9.268869772967448e-06,
|
| 3046 |
+
"loss": 1.8323,
|
| 3047 |
+
"step": 43400
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 0.9067411514570393,
|
| 3051 |
+
"grad_norm": 0.8539250493049622,
|
| 3052 |
+
"learning_rate": 8.942776498847645e-06,
|
| 3053 |
+
"loss": 1.8514,
|
| 3054 |
+
"step": 43500
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 0.9088256138741818,
|
| 3058 |
+
"grad_norm": 0.989700973033905,
|
| 3059 |
+
"learning_rate": 8.617816506222246e-06,
|
| 3060 |
+
"loss": 1.8386,
|
| 3061 |
+
"step": 43600
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 0.9109100762913245,
|
| 3065 |
+
"grad_norm": 0.7901672124862671,
|
| 3066 |
+
"learning_rate": 8.294338133115956e-06,
|
| 3067 |
+
"loss": 1.8395,
|
| 3068 |
+
"step": 43700
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 0.9129945387084671,
|
| 3072 |
+
"grad_norm": 0.9410315155982971,
|
| 3073 |
+
"learning_rate": 7.972688129344382e-06,
|
| 3074 |
+
"loss": 1.8545,
|
| 3075 |
+
"step": 43800
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 0.9150790011256097,
|
| 3079 |
+
"grad_norm": 1.2357829809188843,
|
| 3080 |
+
"learning_rate": 7.653211284818598e-06,
|
| 3081 |
+
"loss": 1.8403,
|
| 3082 |
+
"step": 43900
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 0.9171634635427524,
|
| 3086 |
+
"grad_norm": 0.9447447657585144,
|
| 3087 |
+
"learning_rate": 7.336250059950618e-06,
|
| 3088 |
+
"loss": 1.8304,
|
| 3089 |
+
"step": 44000
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 0.9192479259598949,
|
| 3093 |
+
"grad_norm": 1.0322580337524414,
|
| 3094 |
+
"learning_rate": 7.0221442185559375e-06,
|
| 3095 |
+
"loss": 1.8415,
|
| 3096 |
+
"step": 44100
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 0.9213323883770376,
|
| 3100 |
+
"grad_norm": 0.8802585601806641,
|
| 3101 |
+
"learning_rate": 6.711230463646687e-06,
|
| 3102 |
+
"loss": 1.85,
|
| 3103 |
+
"step": 44200
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 0.9234168507941802,
|
| 3107 |
+
"grad_norm": 1.3486437797546387,
|
| 3108 |
+
"learning_rate": 6.4038420765057915e-06,
|
| 3109 |
+
"loss": 1.8361,
|
| 3110 |
+
"step": 44300
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 0.9255013132113228,
|
| 3114 |
+
"grad_norm": 1.2704123258590698,
|
| 3115 |
+
"learning_rate": 6.100308559428989e-06,
|
| 3116 |
+
"loss": 1.8299,
|
| 3117 |
+
"step": 44400
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 0.9275857756284654,
|
| 3121 |
+
"grad_norm": 0.9107852578163147,
|
| 3122 |
+
"learning_rate": 5.800955282517735e-06,
|
| 3123 |
+
"loss": 1.833,
|
| 3124 |
+
"step": 44500
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 0.929670238045608,
|
| 3128 |
+
"grad_norm": 0.9406431913375854,
|
| 3129 |
+
"learning_rate": 5.50610313490157e-06,
|
| 3130 |
+
"loss": 1.8598,
|
| 3131 |
+
"step": 44600
|
| 3132 |
+
},
|
| 3133 |
+
{
|
| 3134 |
+
"epoch": 0.9317547004627507,
|
| 3135 |
+
"grad_norm": 0.8901939392089844,
|
| 3136 |
+
"learning_rate": 5.2160681807638155e-06,
|
| 3137 |
+
"loss": 1.8485,
|
| 3138 |
+
"step": 44700
|
| 3139 |
+
},
|
| 3140 |
+
{
|
| 3141 |
+
"epoch": 0.9338391628798933,
|
| 3142 |
+
"grad_norm": 0.8982817530632019,
|
| 3143 |
+
"learning_rate": 4.931161320539333e-06,
|
| 3144 |
+
"loss": 1.8423,
|
| 3145 |
+
"step": 44800
|
| 3146 |
+
},
|
| 3147 |
+
{
|
| 3148 |
+
"epoch": 0.9359236252970359,
|
| 3149 |
+
"grad_norm": 1.0009591579437256,
|
| 3150 |
+
"learning_rate": 4.6516879576475195e-06,
|
| 3151 |
+
"loss": 1.8406,
|
| 3152 |
+
"step": 44900
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 0.9380080877141785,
|
| 3156 |
+
"grad_norm": 1.0994175672531128,
|
| 3157 |
+
"learning_rate": 4.377947671117785e-06,
|
| 3158 |
+
"loss": 1.8281,
|
| 3159 |
+
"step": 45000
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 0.9400925501313211,
|
| 3163 |
+
"grad_norm": 0.9065024852752686,
|
| 3164 |
+
"learning_rate": 4.110233894458426e-06,
|
| 3165 |
+
"loss": 1.8417,
|
| 3166 |
+
"step": 45100
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 0.9421770125484638,
|
| 3170 |
+
"grad_norm": 0.9834316372871399,
|
| 3171 |
+
"learning_rate": 3.848833601113141e-06,
|
| 3172 |
+
"loss": 1.8518,
|
| 3173 |
+
"step": 45200
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 0.9442614749656064,
|
| 3177 |
+
"grad_norm": 1.074263572692871,
|
| 3178 |
+
"learning_rate": 3.594026996842356e-06,
|
| 3179 |
+
"loss": 1.8217,
|
| 3180 |
+
"step": 45300
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 0.946345937382749,
|
| 3184 |
+
"grad_norm": 1.2309187650680542,
|
| 3185 |
+
"learning_rate": 3.3460872193591042e-06,
|
| 3186 |
+
"loss": 1.8357,
|
| 3187 |
+
"step": 45400
|
| 3188 |
+
},
|
| 3189 |
+
{
|
| 3190 |
+
"epoch": 0.9484303997998916,
|
| 3191 |
+
"grad_norm": 1.2959349155426025,
|
| 3192 |
+
"learning_rate": 3.1052800455414456e-06,
|
| 3193 |
+
"loss": 1.8349,
|
| 3194 |
+
"step": 45500
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 0.9505148622170342,
|
| 3198 |
+
"grad_norm": 0.8000039458274841,
|
| 3199 |
+
"learning_rate": 2.8718636065352667e-06,
|
| 3200 |
+
"loss": 1.8236,
|
| 3201 |
+
"step": 45600
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 0.9525993246341768,
|
| 3205 |
+
"grad_norm": 1.6720008850097656,
|
| 3206 |
+
"learning_rate": 2.646088111052847e-06,
|
| 3207 |
+
"loss": 1.8372,
|
| 3208 |
+
"step": 45700
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 0.9546837870513195,
|
| 3212 |
+
"grad_norm": 0.9012230038642883,
|
| 3213 |
+
"learning_rate": 2.4281955771638177e-06,
|
| 3214 |
+
"loss": 1.8259,
|
| 3215 |
+
"step": 45800
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 0.9567682494684621,
|
| 3219 |
+
"grad_norm": 1.3057595491409302,
|
| 3220 |
+
"learning_rate": 2.218419572866011e-06,
|
| 3221 |
+
"loss": 1.853,
|
| 3222 |
+
"step": 45900
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 0.9588527118856047,
|
| 3226 |
+
"grad_norm": 0.9006788730621338,
|
| 3227 |
+
"learning_rate": 2.0169849657142825e-06,
|
| 3228 |
+
"loss": 1.8372,
|
| 3229 |
+
"step": 46000
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 0.9609371743027473,
|
| 3233 |
+
"grad_norm": 1.2864965200424194,
|
| 3234 |
+
"learning_rate": 1.8241076817757041e-06,
|
| 3235 |
+
"loss": 1.8409,
|
| 3236 |
+
"step": 46100
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 0.9630216367198899,
|
| 3240 |
+
"grad_norm": 0.9104560613632202,
|
| 3241 |
+
"learning_rate": 1.6399944741694962e-06,
|
| 3242 |
+
"loss": 1.8518,
|
| 3243 |
+
"step": 46200
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 0.9651060991370326,
|
| 3247 |
+
"grad_norm": 0.9562528729438782,
|
| 3248 |
+
"learning_rate": 1.4648427014398336e-06,
|
| 3249 |
+
"loss": 1.8604,
|
| 3250 |
+
"step": 46300
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 0.9671905615541752,
|
| 3254 |
+
"grad_norm": 0.9484128952026367,
|
| 3255 |
+
"learning_rate": 1.2988401159990793e-06,
|
| 3256 |
+
"loss": 1.8423,
|
| 3257 |
+
"step": 46400
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 0.9692750239713178,
|
| 3261 |
+
"grad_norm": 0.9366037845611572,
|
| 3262 |
+
"learning_rate": 1.1421646628682215e-06,
|
| 3263 |
+
"loss": 1.8204,
|
| 3264 |
+
"step": 46500
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 0.9713594863884604,
|
| 3268 |
+
"grad_norm": 0.9006665349006653,
|
| 3269 |
+
"learning_rate": 9.949842889302675e-07,
|
| 3270 |
+
"loss": 1.8443,
|
| 3271 |
+
"step": 46600
|
| 3272 |
+
},
|
| 3273 |
+
{
|
| 3274 |
+
"epoch": 0.973443948805603,
|
| 3275 |
+
"grad_norm": 0.8474847078323364,
|
| 3276 |
+
"learning_rate": 8.574567629010489e-07,
|
| 3277 |
+
"loss": 1.8455,
|
| 3278 |
+
"step": 46700
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 0.9755284112227457,
|
| 3282 |
+
"grad_norm": 0.9436659812927246,
|
| 3283 |
+
"learning_rate": 7.297295062104215e-07,
|
| 3284 |
+
"loss": 1.8402,
|
| 3285 |
+
"step": 46800
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 0.9776128736398882,
|
| 3289 |
+
"grad_norm": 1.5559815168380737,
|
| 3290 |
+
"learning_rate": 6.119394349751628e-07,
|
| 3291 |
+
"loss": 1.8475,
|
| 3292 |
+
"step": 46900
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 0.9796973360570309,
|
| 3296 |
+
"grad_norm": 1.059956431388855,
|
| 3297 |
+
"learning_rate": 5.042128132329205e-07,
|
| 3298 |
+
"loss": 1.8565,
|
| 3299 |
+
"step": 47000
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 0.9817817984741735,
|
| 3303 |
+
"grad_norm": 1.0230424404144287,
|
| 3304 |
+
"learning_rate": 4.0666511759459457e-07,
|
| 3305 |
+
"loss": 1.845,
|
| 3306 |
+
"step": 47100
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 0.9838662608913161,
|
| 3310 |
+
"grad_norm": 0.9298943281173706,
|
| 3311 |
+
"learning_rate": 3.19400913460185e-07,
|
| 3312 |
+
"loss": 1.8235,
|
| 3313 |
+
"step": 47200
|
| 3314 |
+
},
|
| 3315 |
+
{
|
| 3316 |
+
"epoch": 0.9859507233084588,
|
| 3317 |
+
"grad_norm": 1.1216191053390503,
|
| 3318 |
+
"learning_rate": 2.4251374293082885e-07,
|
| 3319 |
+
"loss": 1.8314,
|
| 3320 |
+
"step": 47300
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 0.9880351857256013,
|
| 3324 |
+
"grad_norm": 1.3075478076934814,
|
| 3325 |
+
"learning_rate": 1.7608602453716205e-07,
|
| 3326 |
+
"loss": 1.8308,
|
| 3327 |
+
"step": 47400
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 0.990119648142744,
|
| 3331 |
+
"grad_norm": 0.9057138562202454,
|
| 3332 |
+
"learning_rate": 1.2018896489149333e-07,
|
| 3333 |
+
"loss": 1.8327,
|
| 3334 |
+
"step": 47500
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 0.9922041105598866,
|
| 3338 |
+
"grad_norm": 0.9391676783561707,
|
| 3339 |
+
"learning_rate": 7.488248235850038e-08,
|
| 3340 |
+
"loss": 1.8382,
|
| 3341 |
+
"step": 47600
|
| 3342 |
+
},
|
| 3343 |
+
{
|
| 3344 |
+
"epoch": 0.9942885729770292,
|
| 3345 |
+
"grad_norm": 0.8697664141654968,
|
| 3346 |
+
"learning_rate": 4.0215142826263155e-08,
|
| 3347 |
+
"loss": 1.8641,
|
| 3348 |
+
"step": 47700
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 0.9963730353941719,
|
| 3352 |
+
"grad_norm": 1.663161277770996,
|
| 3353 |
+
"learning_rate": 1.6224107646479747e-08,
|
| 3354 |
+
"loss": 1.8362,
|
| 3355 |
+
"step": 47800
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 0.9984574978113144,
|
| 3359 |
+
"grad_norm": 0.9440004229545593,
|
| 3360 |
+
"learning_rate": 2.935093799688593e-09,
|
| 3361 |
+
"loss": 1.8277,
|
| 3362 |
+
"step": 47900
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 1.0,
|
| 3366 |
+
"step": 47974,
|
| 3367 |
+
"total_flos": 5.824755462670647e+18,
|
| 3368 |
+
"train_loss": 2.213516155517568,
|
| 3369 |
+
"train_runtime": 14249.5377,
|
| 3370 |
+
"train_samples_per_second": 26.934,
|
| 3371 |
+
"train_steps_per_second": 3.367
|
| 3372 |
+
}
|
| 3373 |
+
],
|
| 3374 |
+
"logging_steps": 100,
|
| 3375 |
+
"max_steps": 47974,
|
| 3376 |
+
"num_input_tokens_seen": 0,
|
| 3377 |
+
"num_train_epochs": 1,
|
| 3378 |
+
"save_steps": 500,
|
| 3379 |
+
"stateful_callbacks": {
|
| 3380 |
+
"TrainerControl": {
|
| 3381 |
+
"args": {
|
| 3382 |
+
"should_epoch_stop": false,
|
| 3383 |
+
"should_evaluate": false,
|
| 3384 |
+
"should_log": false,
|
| 3385 |
+
"should_save": false,
|
| 3386 |
+
"should_training_stop": false
|
| 3387 |
+
},
|
| 3388 |
+
"attributes": {}
|
| 3389 |
+
}
|
| 3390 |
+
},
|
| 3391 |
+
"total_flos": 5.824755462670647e+18,
|
| 3392 |
+
"train_batch_size": 1,
|
| 3393 |
+
"trial_name": null,
|
| 3394 |
+
"trial_params": null
|
| 3395 |
+
}
|