Commit
•
ebe708a
1
Parent(s):
5b0f2c8
End of training
Browse files- README.md +5 -0
- all_results.json +16 -0
- eval_results.json +10 -0
- train_results.json +9 -0
- trainer_state.json +3157 -0
README.md
CHANGED
|
@@ -4,6 +4,8 @@ license: mit
|
|
| 4 |
base_model: FacebookAI/roberta-base
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
|
|
|
|
|
|
| 7 |
model-index:
|
| 8 |
- name: populism_multilingual_roberta_base
|
| 9 |
results: []
|
|
@@ -15,6 +17,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
# populism_multilingual_roberta_base
|
| 16 |
|
| 17 |
This model is a fine-tuned version of [FacebookAI/roberta-base](https://huggingface.co/FacebookAI/roberta-base) on an unknown dataset.
|
|
|
|
|
|
|
|
|
|
| 18 |
|
| 19 |
## Model description
|
| 20 |
|
|
|
|
| 4 |
base_model: FacebookAI/roberta-base
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
| 7 |
+
metrics:
|
| 8 |
+
- accuracy
|
| 9 |
model-index:
|
| 10 |
- name: populism_multilingual_roberta_base
|
| 11 |
results: []
|
|
|
|
| 17 |
# populism_multilingual_roberta_base
|
| 18 |
|
| 19 |
This model is a fine-tuned version of [FacebookAI/roberta-base](https://huggingface.co/FacebookAI/roberta-base) on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6781
|
| 22 |
+
- Accuracy: 0.8481
|
| 23 |
|
| 24 |
## Model description
|
| 25 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8481199261869228,
|
| 4 |
+
"eval_loss": 0.678066611289978,
|
| 5 |
+
"eval_runtime": 15544.5436,
|
| 6 |
+
"eval_samples": 2374416,
|
| 7 |
+
"eval_samples_per_second": 152.749,
|
| 8 |
+
"eval_steps_per_second": 1.193,
|
| 9 |
+
"perplexity": 1.9700651461010967,
|
| 10 |
+
"total_flos": 7.501219159976442e+18,
|
| 11 |
+
"train_loss": 0.20371022846941,
|
| 12 |
+
"train_runtime": 62599.538,
|
| 13 |
+
"train_samples": 9497662,
|
| 14 |
+
"train_samples_per_second": 455.163,
|
| 15 |
+
"train_steps_per_second": 3.556
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8481199261869228,
|
| 4 |
+
"eval_loss": 0.678066611289978,
|
| 5 |
+
"eval_runtime": 15544.5436,
|
| 6 |
+
"eval_samples": 2374416,
|
| 7 |
+
"eval_samples_per_second": 152.749,
|
| 8 |
+
"eval_steps_per_second": 1.193,
|
| 9 |
+
"perplexity": 1.9700651461010967
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 7.501219159976442e+18,
|
| 4 |
+
"train_loss": 0.20371022846941,
|
| 5 |
+
"train_runtime": 62599.538,
|
| 6 |
+
"train_samples": 9497662,
|
| 7 |
+
"train_samples_per_second": 455.163,
|
| 8 |
+
"train_steps_per_second": 3.556
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3157 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 222603,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.006738453659654183,
|
| 13 |
+
"grad_norm": 3.9900646209716797,
|
| 14 |
+
"learning_rate": 4.988769243900577e-05,
|
| 15 |
+
"loss": 1.9854,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.013476907319308366,
|
| 20 |
+
"grad_norm": 4.249794006347656,
|
| 21 |
+
"learning_rate": 4.9775384878011525e-05,
|
| 22 |
+
"loss": 1.8451,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.020215360978962548,
|
| 27 |
+
"grad_norm": 4.1350417137146,
|
| 28 |
+
"learning_rate": 4.966307731701729e-05,
|
| 29 |
+
"loss": 1.7752,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.026953814638616732,
|
| 34 |
+
"grad_norm": 3.874133825302124,
|
| 35 |
+
"learning_rate": 4.955076975602306e-05,
|
| 36 |
+
"loss": 1.7215,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.03369226829827091,
|
| 41 |
+
"grad_norm": 3.557955026626587,
|
| 42 |
+
"learning_rate": 4.943846219502882e-05,
|
| 43 |
+
"loss": 1.6764,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.040430721957925096,
|
| 48 |
+
"grad_norm": 3.978665590286255,
|
| 49 |
+
"learning_rate": 4.932615463403458e-05,
|
| 50 |
+
"loss": 1.6395,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.04716917561757928,
|
| 55 |
+
"grad_norm": 3.99066162109375,
|
| 56 |
+
"learning_rate": 4.921384707304035e-05,
|
| 57 |
+
"loss": 1.6005,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.053907629277233464,
|
| 62 |
+
"grad_norm": 3.6204793453216553,
|
| 63 |
+
"learning_rate": 4.910153951204611e-05,
|
| 64 |
+
"loss": 1.5745,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.06064608293688764,
|
| 69 |
+
"grad_norm": 3.5526976585388184,
|
| 70 |
+
"learning_rate": 4.898923195105187e-05,
|
| 71 |
+
"loss": 1.5477,
|
| 72 |
+
"step": 4500
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.06738453659654182,
|
| 76 |
+
"grad_norm": 3.432760238647461,
|
| 77 |
+
"learning_rate": 4.887692439005764e-05,
|
| 78 |
+
"loss": 1.5235,
|
| 79 |
+
"step": 5000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.07412299025619601,
|
| 83 |
+
"grad_norm": 3.3595640659332275,
|
| 84 |
+
"learning_rate": 4.87646168290634e-05,
|
| 85 |
+
"loss": 1.5069,
|
| 86 |
+
"step": 5500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.08086144391585019,
|
| 90 |
+
"grad_norm": 3.6151161193847656,
|
| 91 |
+
"learning_rate": 4.865230926806917e-05,
|
| 92 |
+
"loss": 1.4855,
|
| 93 |
+
"step": 6000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.08759989757550438,
|
| 97 |
+
"grad_norm": 3.4927330017089844,
|
| 98 |
+
"learning_rate": 4.854000170707493e-05,
|
| 99 |
+
"loss": 1.4636,
|
| 100 |
+
"step": 6500
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.09433835123515856,
|
| 104 |
+
"grad_norm": 5.184919357299805,
|
| 105 |
+
"learning_rate": 4.842769414608069e-05,
|
| 106 |
+
"loss": 1.455,
|
| 107 |
+
"step": 7000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.10107680489481274,
|
| 111 |
+
"grad_norm": 3.0606026649475098,
|
| 112 |
+
"learning_rate": 4.831538658508646e-05,
|
| 113 |
+
"loss": 1.437,
|
| 114 |
+
"step": 7500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.10781525855446693,
|
| 118 |
+
"grad_norm": 3.1713528633117676,
|
| 119 |
+
"learning_rate": 4.820307902409222e-05,
|
| 120 |
+
"loss": 1.4222,
|
| 121 |
+
"step": 8000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.1145537122141211,
|
| 125 |
+
"grad_norm": 3.203773260116577,
|
| 126 |
+
"learning_rate": 4.809077146309798e-05,
|
| 127 |
+
"loss": 1.4091,
|
| 128 |
+
"step": 8500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.12129216587377528,
|
| 132 |
+
"grad_norm": 3.338106393814087,
|
| 133 |
+
"learning_rate": 4.797846390210375e-05,
|
| 134 |
+
"loss": 1.3962,
|
| 135 |
+
"step": 9000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.12803061953342948,
|
| 139 |
+
"grad_norm": 3.13653302192688,
|
| 140 |
+
"learning_rate": 4.786615634110951e-05,
|
| 141 |
+
"loss": 1.386,
|
| 142 |
+
"step": 9500
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.13476907319308365,
|
| 146 |
+
"grad_norm": 3.3359720706939697,
|
| 147 |
+
"learning_rate": 4.775384878011527e-05,
|
| 148 |
+
"loss": 1.3716,
|
| 149 |
+
"step": 10000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.14150752685273785,
|
| 153 |
+
"grad_norm": 3.115161418914795,
|
| 154 |
+
"learning_rate": 4.764154121912104e-05,
|
| 155 |
+
"loss": 1.363,
|
| 156 |
+
"step": 10500
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.14824598051239202,
|
| 160 |
+
"grad_norm": 3.1449685096740723,
|
| 161 |
+
"learning_rate": 4.75292336581268e-05,
|
| 162 |
+
"loss": 1.3504,
|
| 163 |
+
"step": 11000
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.1549844341720462,
|
| 167 |
+
"grad_norm": 3.3429105281829834,
|
| 168 |
+
"learning_rate": 4.741692609713256e-05,
|
| 169 |
+
"loss": 1.341,
|
| 170 |
+
"step": 11500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.16172288783170038,
|
| 174 |
+
"grad_norm": 3.8695995807647705,
|
| 175 |
+
"learning_rate": 4.730461853613833e-05,
|
| 176 |
+
"loss": 1.3312,
|
| 177 |
+
"step": 12000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.16846134149135455,
|
| 181 |
+
"grad_norm": 3.238085985183716,
|
| 182 |
+
"learning_rate": 4.719231097514409e-05,
|
| 183 |
+
"loss": 1.3144,
|
| 184 |
+
"step": 12500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.17519979515100875,
|
| 188 |
+
"grad_norm": 2.9137234687805176,
|
| 189 |
+
"learning_rate": 4.708000341414986e-05,
|
| 190 |
+
"loss": 1.3137,
|
| 191 |
+
"step": 13000
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.18193824881066292,
|
| 195 |
+
"grad_norm": 3.0767650604248047,
|
| 196 |
+
"learning_rate": 4.696769585315562e-05,
|
| 197 |
+
"loss": 1.2981,
|
| 198 |
+
"step": 13500
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.18867670247031712,
|
| 202 |
+
"grad_norm": 3.804330825805664,
|
| 203 |
+
"learning_rate": 4.685538829216138e-05,
|
| 204 |
+
"loss": 1.294,
|
| 205 |
+
"step": 14000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.1954151561299713,
|
| 209 |
+
"grad_norm": 3.352318286895752,
|
| 210 |
+
"learning_rate": 4.674308073116715e-05,
|
| 211 |
+
"loss": 1.2855,
|
| 212 |
+
"step": 14500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.2021536097896255,
|
| 216 |
+
"grad_norm": 2.844233274459839,
|
| 217 |
+
"learning_rate": 4.663077317017291e-05,
|
| 218 |
+
"loss": 1.2776,
|
| 219 |
+
"step": 15000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.20889206344927966,
|
| 223 |
+
"grad_norm": 3.068960666656494,
|
| 224 |
+
"learning_rate": 4.651846560917867e-05,
|
| 225 |
+
"loss": 1.2656,
|
| 226 |
+
"step": 15500
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.21563051710893386,
|
| 230 |
+
"grad_norm": 3.283202648162842,
|
| 231 |
+
"learning_rate": 4.640615804818444e-05,
|
| 232 |
+
"loss": 1.2613,
|
| 233 |
+
"step": 16000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.22236897076858803,
|
| 237 |
+
"grad_norm": 3.0281543731689453,
|
| 238 |
+
"learning_rate": 4.629385048719021e-05,
|
| 239 |
+
"loss": 1.2607,
|
| 240 |
+
"step": 16500
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.2291074244282422,
|
| 244 |
+
"grad_norm": 3.088404417037964,
|
| 245 |
+
"learning_rate": 4.618154292619596e-05,
|
| 246 |
+
"loss": 1.2531,
|
| 247 |
+
"step": 17000
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.2358458780878964,
|
| 251 |
+
"grad_norm": 2.955118417739868,
|
| 252 |
+
"learning_rate": 4.606923536520173e-05,
|
| 253 |
+
"loss": 1.2515,
|
| 254 |
+
"step": 17500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.24258433174755056,
|
| 258 |
+
"grad_norm": 3.179586172103882,
|
| 259 |
+
"learning_rate": 4.59569278042075e-05,
|
| 260 |
+
"loss": 1.2312,
|
| 261 |
+
"step": 18000
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.24932278540720476,
|
| 265 |
+
"grad_norm": 3.088040590286255,
|
| 266 |
+
"learning_rate": 4.584462024321325e-05,
|
| 267 |
+
"loss": 1.2278,
|
| 268 |
+
"step": 18500
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.25606123906685896,
|
| 272 |
+
"grad_norm": 3.2045764923095703,
|
| 273 |
+
"learning_rate": 4.573231268221902e-05,
|
| 274 |
+
"loss": 1.2248,
|
| 275 |
+
"step": 19000
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.2627996927265131,
|
| 279 |
+
"grad_norm": 3.1761693954467773,
|
| 280 |
+
"learning_rate": 4.562000512122479e-05,
|
| 281 |
+
"loss": 1.2165,
|
| 282 |
+
"step": 19500
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.2695381463861673,
|
| 286 |
+
"grad_norm": 3.1050639152526855,
|
| 287 |
+
"learning_rate": 4.550769756023054e-05,
|
| 288 |
+
"loss": 1.2115,
|
| 289 |
+
"step": 20000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.2762766000458215,
|
| 293 |
+
"grad_norm": 3.4235286712646484,
|
| 294 |
+
"learning_rate": 4.539538999923631e-05,
|
| 295 |
+
"loss": 1.2124,
|
| 296 |
+
"step": 20500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.2830150537054757,
|
| 300 |
+
"grad_norm": 3.28312611579895,
|
| 301 |
+
"learning_rate": 4.528308243824208e-05,
|
| 302 |
+
"loss": 1.1959,
|
| 303 |
+
"step": 21000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.28975350736512984,
|
| 307 |
+
"grad_norm": 3.0334129333496094,
|
| 308 |
+
"learning_rate": 4.517077487724784e-05,
|
| 309 |
+
"loss": 1.2004,
|
| 310 |
+
"step": 21500
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.29649196102478403,
|
| 314 |
+
"grad_norm": 2.892882823944092,
|
| 315 |
+
"learning_rate": 4.50584673162536e-05,
|
| 316 |
+
"loss": 1.1927,
|
| 317 |
+
"step": 22000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.30323041468443823,
|
| 321 |
+
"grad_norm": 3.4390981197357178,
|
| 322 |
+
"learning_rate": 4.494615975525937e-05,
|
| 323 |
+
"loss": 1.1876,
|
| 324 |
+
"step": 22500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.3099688683440924,
|
| 328 |
+
"grad_norm": 3.093661308288574,
|
| 329 |
+
"learning_rate": 4.483385219426513e-05,
|
| 330 |
+
"loss": 1.1845,
|
| 331 |
+
"step": 23000
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.31670732200374657,
|
| 335 |
+
"grad_norm": 3.129610538482666,
|
| 336 |
+
"learning_rate": 4.472154463327089e-05,
|
| 337 |
+
"loss": 1.1798,
|
| 338 |
+
"step": 23500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.32344577566340077,
|
| 342 |
+
"grad_norm": 2.9947609901428223,
|
| 343 |
+
"learning_rate": 4.460923707227666e-05,
|
| 344 |
+
"loss": 1.1788,
|
| 345 |
+
"step": 24000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.33018422932305497,
|
| 349 |
+
"grad_norm": 3.019880771636963,
|
| 350 |
+
"learning_rate": 4.449692951128242e-05,
|
| 351 |
+
"loss": 1.166,
|
| 352 |
+
"step": 24500
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.3369226829827091,
|
| 356 |
+
"grad_norm": 2.9221296310424805,
|
| 357 |
+
"learning_rate": 4.438462195028819e-05,
|
| 358 |
+
"loss": 1.1609,
|
| 359 |
+
"step": 25000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.3436611366423633,
|
| 363 |
+
"grad_norm": 3.045243740081787,
|
| 364 |
+
"learning_rate": 4.427231438929395e-05,
|
| 365 |
+
"loss": 1.1621,
|
| 366 |
+
"step": 25500
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.3503995903020175,
|
| 370 |
+
"grad_norm": 3.090078592300415,
|
| 371 |
+
"learning_rate": 4.416000682829971e-05,
|
| 372 |
+
"loss": 1.1511,
|
| 373 |
+
"step": 26000
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.3571380439616717,
|
| 377 |
+
"grad_norm": 2.8891849517822266,
|
| 378 |
+
"learning_rate": 4.404769926730548e-05,
|
| 379 |
+
"loss": 1.1525,
|
| 380 |
+
"step": 26500
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.36387649762132585,
|
| 384 |
+
"grad_norm": 2.9999043941497803,
|
| 385 |
+
"learning_rate": 4.393539170631124e-05,
|
| 386 |
+
"loss": 1.1466,
|
| 387 |
+
"step": 27000
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.37061495128098004,
|
| 391 |
+
"grad_norm": 2.739863872528076,
|
| 392 |
+
"learning_rate": 4.3823084145317e-05,
|
| 393 |
+
"loss": 1.139,
|
| 394 |
+
"step": 27500
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.37735340494063424,
|
| 398 |
+
"grad_norm": 3.058260679244995,
|
| 399 |
+
"learning_rate": 4.371077658432277e-05,
|
| 400 |
+
"loss": 1.1386,
|
| 401 |
+
"step": 28000
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.3840918586002884,
|
| 405 |
+
"grad_norm": 3.324040651321411,
|
| 406 |
+
"learning_rate": 4.359846902332853e-05,
|
| 407 |
+
"loss": 1.13,
|
| 408 |
+
"step": 28500
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.3908303122599426,
|
| 412 |
+
"grad_norm": 2.7937138080596924,
|
| 413 |
+
"learning_rate": 4.348616146233429e-05,
|
| 414 |
+
"loss": 1.1325,
|
| 415 |
+
"step": 29000
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.3975687659195968,
|
| 419 |
+
"grad_norm": 2.886350631713867,
|
| 420 |
+
"learning_rate": 4.337385390134006e-05,
|
| 421 |
+
"loss": 1.1242,
|
| 422 |
+
"step": 29500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.404307219579251,
|
| 426 |
+
"grad_norm": 3.051133871078491,
|
| 427 |
+
"learning_rate": 4.326154634034582e-05,
|
| 428 |
+
"loss": 1.1206,
|
| 429 |
+
"step": 30000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.4110456732389051,
|
| 433 |
+
"grad_norm": 2.869065046310425,
|
| 434 |
+
"learning_rate": 4.314923877935158e-05,
|
| 435 |
+
"loss": 1.1213,
|
| 436 |
+
"step": 30500
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.4177841268985593,
|
| 440 |
+
"grad_norm": 3.144162178039551,
|
| 441 |
+
"learning_rate": 4.303693121835735e-05,
|
| 442 |
+
"loss": 1.1197,
|
| 443 |
+
"step": 31000
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.4245225805582135,
|
| 447 |
+
"grad_norm": 2.7358269691467285,
|
| 448 |
+
"learning_rate": 4.292462365736311e-05,
|
| 449 |
+
"loss": 1.1139,
|
| 450 |
+
"step": 31500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.4312610342178677,
|
| 454 |
+
"grad_norm": 2.651580572128296,
|
| 455 |
+
"learning_rate": 4.281231609636888e-05,
|
| 456 |
+
"loss": 1.1104,
|
| 457 |
+
"step": 32000
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.43799948787752185,
|
| 461 |
+
"grad_norm": 2.9672605991363525,
|
| 462 |
+
"learning_rate": 4.270000853537464e-05,
|
| 463 |
+
"loss": 1.1067,
|
| 464 |
+
"step": 32500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.44473794153717605,
|
| 468 |
+
"grad_norm": 2.8731400966644287,
|
| 469 |
+
"learning_rate": 4.25877009743804e-05,
|
| 470 |
+
"loss": 1.1071,
|
| 471 |
+
"step": 33000
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.45147639519683025,
|
| 475 |
+
"grad_norm": 2.9311606884002686,
|
| 476 |
+
"learning_rate": 4.247539341338617e-05,
|
| 477 |
+
"loss": 1.1054,
|
| 478 |
+
"step": 33500
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.4582148488564844,
|
| 482 |
+
"grad_norm": 3.236234188079834,
|
| 483 |
+
"learning_rate": 4.236308585239193e-05,
|
| 484 |
+
"loss": 1.1021,
|
| 485 |
+
"step": 34000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.4649533025161386,
|
| 489 |
+
"grad_norm": 2.785691738128662,
|
| 490 |
+
"learning_rate": 4.225077829139769e-05,
|
| 491 |
+
"loss": 1.0948,
|
| 492 |
+
"step": 34500
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.4716917561757928,
|
| 496 |
+
"grad_norm": 3.1143743991851807,
|
| 497 |
+
"learning_rate": 4.213847073040346e-05,
|
| 498 |
+
"loss": 1.0935,
|
| 499 |
+
"step": 35000
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.478430209835447,
|
| 503 |
+
"grad_norm": 3.0961947441101074,
|
| 504 |
+
"learning_rate": 4.2026163169409225e-05,
|
| 505 |
+
"loss": 1.0867,
|
| 506 |
+
"step": 35500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.4851686634951011,
|
| 510 |
+
"grad_norm": 2.7408347129821777,
|
| 511 |
+
"learning_rate": 4.191385560841498e-05,
|
| 512 |
+
"loss": 1.089,
|
| 513 |
+
"step": 36000
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.4919071171547553,
|
| 517 |
+
"grad_norm": 3.244724750518799,
|
| 518 |
+
"learning_rate": 4.180154804742075e-05,
|
| 519 |
+
"loss": 1.0867,
|
| 520 |
+
"step": 36500
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.4986455708144095,
|
| 524 |
+
"grad_norm": 2.8332059383392334,
|
| 525 |
+
"learning_rate": 4.168924048642651e-05,
|
| 526 |
+
"loss": 1.0763,
|
| 527 |
+
"step": 37000
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.5053840244740637,
|
| 531 |
+
"grad_norm": 2.896265745162964,
|
| 532 |
+
"learning_rate": 4.157693292543227e-05,
|
| 533 |
+
"loss": 1.0696,
|
| 534 |
+
"step": 37500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.5121224781337179,
|
| 538 |
+
"grad_norm": 2.8772451877593994,
|
| 539 |
+
"learning_rate": 4.146462536443804e-05,
|
| 540 |
+
"loss": 1.0727,
|
| 541 |
+
"step": 38000
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.518860931793372,
|
| 545 |
+
"grad_norm": 2.5283446311950684,
|
| 546 |
+
"learning_rate": 4.13523178034438e-05,
|
| 547 |
+
"loss": 1.068,
|
| 548 |
+
"step": 38500
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.5255993854530262,
|
| 552 |
+
"grad_norm": 2.669386148452759,
|
| 553 |
+
"learning_rate": 4.1240010242449567e-05,
|
| 554 |
+
"loss": 1.0738,
|
| 555 |
+
"step": 39000
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.5323378391126804,
|
| 559 |
+
"grad_norm": 2.9089040756225586,
|
| 560 |
+
"learning_rate": 4.112770268145533e-05,
|
| 561 |
+
"loss": 1.0677,
|
| 562 |
+
"step": 39500
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.5390762927723346,
|
| 566 |
+
"grad_norm": 2.889467716217041,
|
| 567 |
+
"learning_rate": 4.101539512046109e-05,
|
| 568 |
+
"loss": 1.0613,
|
| 569 |
+
"step": 40000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.5458147464319888,
|
| 573 |
+
"grad_norm": 2.6893374919891357,
|
| 574 |
+
"learning_rate": 4.090308755946686e-05,
|
| 575 |
+
"loss": 1.0644,
|
| 576 |
+
"step": 40500
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.552553200091643,
|
| 580 |
+
"grad_norm": 2.7254648208618164,
|
| 581 |
+
"learning_rate": 4.079077999847262e-05,
|
| 582 |
+
"loss": 1.0593,
|
| 583 |
+
"step": 41000
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.5592916537512972,
|
| 587 |
+
"grad_norm": 2.8278279304504395,
|
| 588 |
+
"learning_rate": 4.067847243747838e-05,
|
| 589 |
+
"loss": 1.0534,
|
| 590 |
+
"step": 41500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.5660301074109514,
|
| 594 |
+
"grad_norm": 2.885976791381836,
|
| 595 |
+
"learning_rate": 4.056616487648415e-05,
|
| 596 |
+
"loss": 1.0495,
|
| 597 |
+
"step": 42000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.5727685610706055,
|
| 601 |
+
"grad_norm": 2.6645610332489014,
|
| 602 |
+
"learning_rate": 4.0453857315489915e-05,
|
| 603 |
+
"loss": 1.0502,
|
| 604 |
+
"step": 42500
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.5795070147302597,
|
| 608 |
+
"grad_norm": 2.8876936435699463,
|
| 609 |
+
"learning_rate": 4.034154975449567e-05,
|
| 610 |
+
"loss": 1.0457,
|
| 611 |
+
"step": 43000
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.5862454683899139,
|
| 615 |
+
"grad_norm": 2.9418463706970215,
|
| 616 |
+
"learning_rate": 4.022924219350144e-05,
|
| 617 |
+
"loss": 1.0459,
|
| 618 |
+
"step": 43500
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.5929839220495681,
|
| 622 |
+
"grad_norm": 2.9865574836730957,
|
| 623 |
+
"learning_rate": 4.0116934632507205e-05,
|
| 624 |
+
"loss": 1.0379,
|
| 625 |
+
"step": 44000
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.5997223757092223,
|
| 629 |
+
"grad_norm": 2.832772731781006,
|
| 630 |
+
"learning_rate": 4.000462707151296e-05,
|
| 631 |
+
"loss": 1.0439,
|
| 632 |
+
"step": 44500
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.6064608293688765,
|
| 636 |
+
"grad_norm": 2.743056535720825,
|
| 637 |
+
"learning_rate": 3.989231951051873e-05,
|
| 638 |
+
"loss": 1.0325,
|
| 639 |
+
"step": 45000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.6131992830285307,
|
| 643 |
+
"grad_norm": 2.9779651165008545,
|
| 644 |
+
"learning_rate": 3.9780011949524495e-05,
|
| 645 |
+
"loss": 1.0404,
|
| 646 |
+
"step": 45500
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.6199377366881847,
|
| 650 |
+
"grad_norm": 2.8546512126922607,
|
| 651 |
+
"learning_rate": 3.9667704388530256e-05,
|
| 652 |
+
"loss": 1.0345,
|
| 653 |
+
"step": 46000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.626676190347839,
|
| 657 |
+
"grad_norm": 2.870129108428955,
|
| 658 |
+
"learning_rate": 3.955539682753602e-05,
|
| 659 |
+
"loss": 1.0362,
|
| 660 |
+
"step": 46500
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.6334146440074931,
|
| 664 |
+
"grad_norm": 5.683381080627441,
|
| 665 |
+
"learning_rate": 3.9443089266541785e-05,
|
| 666 |
+
"loss": 1.0301,
|
| 667 |
+
"step": 47000
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.6401530976671473,
|
| 671 |
+
"grad_norm": 3.0040597915649414,
|
| 672 |
+
"learning_rate": 3.9330781705547546e-05,
|
| 673 |
+
"loss": 1.0259,
|
| 674 |
+
"step": 47500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.6468915513268015,
|
| 678 |
+
"grad_norm": 3.0165131092071533,
|
| 679 |
+
"learning_rate": 3.921847414455331e-05,
|
| 680 |
+
"loss": 1.0256,
|
| 681 |
+
"step": 48000
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.6536300049864557,
|
| 685 |
+
"grad_norm": 2.627032995223999,
|
| 686 |
+
"learning_rate": 3.9106166583559075e-05,
|
| 687 |
+
"loss": 1.023,
|
| 688 |
+
"step": 48500
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.6603684586461099,
|
| 692 |
+
"grad_norm": 2.998840808868408,
|
| 693 |
+
"learning_rate": 3.8993859022564836e-05,
|
| 694 |
+
"loss": 1.0243,
|
| 695 |
+
"step": 49000
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.667106912305764,
|
| 699 |
+
"grad_norm": 2.4952380657196045,
|
| 700 |
+
"learning_rate": 3.8881551461570604e-05,
|
| 701 |
+
"loss": 1.0224,
|
| 702 |
+
"step": 49500
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.6738453659654182,
|
| 706 |
+
"grad_norm": 2.69301176071167,
|
| 707 |
+
"learning_rate": 3.8769243900576365e-05,
|
| 708 |
+
"loss": 1.0175,
|
| 709 |
+
"step": 50000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.6805838196250724,
|
| 713 |
+
"grad_norm": 2.8481318950653076,
|
| 714 |
+
"learning_rate": 3.8656936339582127e-05,
|
| 715 |
+
"loss": 1.0116,
|
| 716 |
+
"step": 50500
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.6873222732847266,
|
| 720 |
+
"grad_norm": 2.9201319217681885,
|
| 721 |
+
"learning_rate": 3.8544628778587894e-05,
|
| 722 |
+
"loss": 1.0158,
|
| 723 |
+
"step": 51000
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.6940607269443808,
|
| 727 |
+
"grad_norm": 2.8624720573425293,
|
| 728 |
+
"learning_rate": 3.8432321217593656e-05,
|
| 729 |
+
"loss": 1.0086,
|
| 730 |
+
"step": 51500
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.700799180604035,
|
| 734 |
+
"grad_norm": 2.8680732250213623,
|
| 735 |
+
"learning_rate": 3.832001365659942e-05,
|
| 736 |
+
"loss": 1.0048,
|
| 737 |
+
"step": 52000
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.7075376342636892,
|
| 741 |
+
"grad_norm": 3.0876636505126953,
|
| 742 |
+
"learning_rate": 3.8207706095605185e-05,
|
| 743 |
+
"loss": 1.0032,
|
| 744 |
+
"step": 52500
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.7142760879233434,
|
| 748 |
+
"grad_norm": 2.580636501312256,
|
| 749 |
+
"learning_rate": 3.8095398534610946e-05,
|
| 750 |
+
"loss": 1.0071,
|
| 751 |
+
"step": 53000
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.7210145415829975,
|
| 755 |
+
"grad_norm": 2.7390966415405273,
|
| 756 |
+
"learning_rate": 3.798309097361671e-05,
|
| 757 |
+
"loss": 1.0009,
|
| 758 |
+
"step": 53500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.7277529952426517,
|
| 762 |
+
"grad_norm": 2.596480369567871,
|
| 763 |
+
"learning_rate": 3.7870783412622475e-05,
|
| 764 |
+
"loss": 1.0047,
|
| 765 |
+
"step": 54000
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.7344914489023059,
|
| 769 |
+
"grad_norm": 2.96301007270813,
|
| 770 |
+
"learning_rate": 3.7758475851628236e-05,
|
| 771 |
+
"loss": 1.0043,
|
| 772 |
+
"step": 54500
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.7412299025619601,
|
| 776 |
+
"grad_norm": 2.991548776626587,
|
| 777 |
+
"learning_rate": 3.7646168290634e-05,
|
| 778 |
+
"loss": 0.999,
|
| 779 |
+
"step": 55000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.7479683562216143,
|
| 783 |
+
"grad_norm": 3.175384521484375,
|
| 784 |
+
"learning_rate": 3.7533860729639765e-05,
|
| 785 |
+
"loss": 0.9908,
|
| 786 |
+
"step": 55500
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.7547068098812685,
|
| 790 |
+
"grad_norm": 2.9244823455810547,
|
| 791 |
+
"learning_rate": 3.7421553168645526e-05,
|
| 792 |
+
"loss": 0.9951,
|
| 793 |
+
"step": 56000
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.7614452635409227,
|
| 797 |
+
"grad_norm": 3.0108466148376465,
|
| 798 |
+
"learning_rate": 3.730924560765129e-05,
|
| 799 |
+
"loss": 0.9923,
|
| 800 |
+
"step": 56500
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.7681837172005768,
|
| 804 |
+
"grad_norm": 2.9138221740722656,
|
| 805 |
+
"learning_rate": 3.7196938046657055e-05,
|
| 806 |
+
"loss": 0.9884,
|
| 807 |
+
"step": 57000
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.774922170860231,
|
| 811 |
+
"grad_norm": 2.9478752613067627,
|
| 812 |
+
"learning_rate": 3.7084630485662816e-05,
|
| 813 |
+
"loss": 0.9882,
|
| 814 |
+
"step": 57500
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.7816606245198852,
|
| 818 |
+
"grad_norm": 2.4768459796905518,
|
| 819 |
+
"learning_rate": 3.6972322924668584e-05,
|
| 820 |
+
"loss": 0.9892,
|
| 821 |
+
"step": 58000
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.7883990781795394,
|
| 825 |
+
"grad_norm": 2.681396007537842,
|
| 826 |
+
"learning_rate": 3.6860015363674345e-05,
|
| 827 |
+
"loss": 0.9829,
|
| 828 |
+
"step": 58500
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.7951375318391936,
|
| 832 |
+
"grad_norm": 2.6033999919891357,
|
| 833 |
+
"learning_rate": 3.6747707802680106e-05,
|
| 834 |
+
"loss": 0.9858,
|
| 835 |
+
"step": 59000
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.8018759854988478,
|
| 839 |
+
"grad_norm": 2.896912097930908,
|
| 840 |
+
"learning_rate": 3.6635400241685874e-05,
|
| 841 |
+
"loss": 0.9824,
|
| 842 |
+
"step": 59500
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.808614439158502,
|
| 846 |
+
"grad_norm": 2.609686851501465,
|
| 847 |
+
"learning_rate": 3.6523092680691635e-05,
|
| 848 |
+
"loss": 0.9765,
|
| 849 |
+
"step": 60000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.815352892818156,
|
| 853 |
+
"grad_norm": 2.8086376190185547,
|
| 854 |
+
"learning_rate": 3.6410785119697396e-05,
|
| 855 |
+
"loss": 0.9784,
|
| 856 |
+
"step": 60500
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.8220913464778102,
|
| 860 |
+
"grad_norm": 2.676877498626709,
|
| 861 |
+
"learning_rate": 3.6298477558703164e-05,
|
| 862 |
+
"loss": 0.978,
|
| 863 |
+
"step": 61000
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.8288298001374644,
|
| 867 |
+
"grad_norm": 2.8640575408935547,
|
| 868 |
+
"learning_rate": 3.618616999770893e-05,
|
| 869 |
+
"loss": 0.973,
|
| 870 |
+
"step": 61500
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.8355682537971186,
|
| 874 |
+
"grad_norm": 2.6603965759277344,
|
| 875 |
+
"learning_rate": 3.6073862436714687e-05,
|
| 876 |
+
"loss": 0.9737,
|
| 877 |
+
"step": 62000
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.8423067074567728,
|
| 881 |
+
"grad_norm": 2.7209627628326416,
|
| 882 |
+
"learning_rate": 3.5961554875720454e-05,
|
| 883 |
+
"loss": 0.9725,
|
| 884 |
+
"step": 62500
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.849045161116427,
|
| 888 |
+
"grad_norm": 2.7673871517181396,
|
| 889 |
+
"learning_rate": 3.584924731472622e-05,
|
| 890 |
+
"loss": 0.9707,
|
| 891 |
+
"step": 63000
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.8557836147760812,
|
| 895 |
+
"grad_norm": 2.5891125202178955,
|
| 896 |
+
"learning_rate": 3.573693975373198e-05,
|
| 897 |
+
"loss": 0.9665,
|
| 898 |
+
"step": 63500
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.8625220684357354,
|
| 902 |
+
"grad_norm": 2.7834441661834717,
|
| 903 |
+
"learning_rate": 3.5624632192737745e-05,
|
| 904 |
+
"loss": 0.9696,
|
| 905 |
+
"step": 64000
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.8692605220953895,
|
| 909 |
+
"grad_norm": 2.835066318511963,
|
| 910 |
+
"learning_rate": 3.551232463174351e-05,
|
| 911 |
+
"loss": 0.9766,
|
| 912 |
+
"step": 64500
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.8759989757550437,
|
| 916 |
+
"grad_norm": 3.202230453491211,
|
| 917 |
+
"learning_rate": 3.5400017070749274e-05,
|
| 918 |
+
"loss": 0.9654,
|
| 919 |
+
"step": 65000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.8827374294146979,
|
| 923 |
+
"grad_norm": 2.8752474784851074,
|
| 924 |
+
"learning_rate": 3.5287709509755035e-05,
|
| 925 |
+
"loss": 0.9608,
|
| 926 |
+
"step": 65500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.8894758830743521,
|
| 930 |
+
"grad_norm": 2.9779469966888428,
|
| 931 |
+
"learning_rate": 3.51754019487608e-05,
|
| 932 |
+
"loss": 0.9605,
|
| 933 |
+
"step": 66000
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.8962143367340063,
|
| 937 |
+
"grad_norm": 2.537226438522339,
|
| 938 |
+
"learning_rate": 3.5063094387766564e-05,
|
| 939 |
+
"loss": 0.955,
|
| 940 |
+
"step": 66500
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.9029527903936605,
|
| 944 |
+
"grad_norm": 2.680373191833496,
|
| 945 |
+
"learning_rate": 3.4950786826772325e-05,
|
| 946 |
+
"loss": 0.9568,
|
| 947 |
+
"step": 67000
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.9096912440533147,
|
| 951 |
+
"grad_norm": 2.559218406677246,
|
| 952 |
+
"learning_rate": 3.483847926577809e-05,
|
| 953 |
+
"loss": 0.9568,
|
| 954 |
+
"step": 67500
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.9164296977129688,
|
| 958 |
+
"grad_norm": 2.6627488136291504,
|
| 959 |
+
"learning_rate": 3.4726171704783854e-05,
|
| 960 |
+
"loss": 0.9508,
|
| 961 |
+
"step": 68000
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.923168151372623,
|
| 965 |
+
"grad_norm": 2.8378233909606934,
|
| 966 |
+
"learning_rate": 3.461386414378962e-05,
|
| 967 |
+
"loss": 0.9573,
|
| 968 |
+
"step": 68500
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.9299066050322772,
|
| 972 |
+
"grad_norm": 2.688589572906494,
|
| 973 |
+
"learning_rate": 3.450155658279538e-05,
|
| 974 |
+
"loss": 0.9505,
|
| 975 |
+
"step": 69000
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.9366450586919314,
|
| 979 |
+
"grad_norm": 2.9189035892486572,
|
| 980 |
+
"learning_rate": 3.4389249021801144e-05,
|
| 981 |
+
"loss": 0.9519,
|
| 982 |
+
"step": 69500
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.9433835123515856,
|
| 986 |
+
"grad_norm": 2.8582780361175537,
|
| 987 |
+
"learning_rate": 3.427694146080691e-05,
|
| 988 |
+
"loss": 0.9464,
|
| 989 |
+
"step": 70000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.9501219660112398,
|
| 993 |
+
"grad_norm": 2.728365659713745,
|
| 994 |
+
"learning_rate": 3.416463389981267e-05,
|
| 995 |
+
"loss": 0.9487,
|
| 996 |
+
"step": 70500
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.956860419670894,
|
| 1000 |
+
"grad_norm": 2.6869001388549805,
|
| 1001 |
+
"learning_rate": 3.4052326338818434e-05,
|
| 1002 |
+
"loss": 0.9454,
|
| 1003 |
+
"step": 71000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.9635988733305481,
|
| 1007 |
+
"grad_norm": 2.6964008808135986,
|
| 1008 |
+
"learning_rate": 3.39400187778242e-05,
|
| 1009 |
+
"loss": 0.9445,
|
| 1010 |
+
"step": 71500
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.9703373269902023,
|
| 1014 |
+
"grad_norm": 2.9616520404815674,
|
| 1015 |
+
"learning_rate": 3.382771121682996e-05,
|
| 1016 |
+
"loss": 0.9452,
|
| 1017 |
+
"step": 72000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.9770757806498565,
|
| 1021 |
+
"grad_norm": 2.671269178390503,
|
| 1022 |
+
"learning_rate": 3.3715403655835724e-05,
|
| 1023 |
+
"loss": 0.9431,
|
| 1024 |
+
"step": 72500
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.9838142343095106,
|
| 1028 |
+
"grad_norm": 2.4057188034057617,
|
| 1029 |
+
"learning_rate": 3.360309609484149e-05,
|
| 1030 |
+
"loss": 0.9404,
|
| 1031 |
+
"step": 73000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.9905526879691648,
|
| 1035 |
+
"grad_norm": 2.8451008796691895,
|
| 1036 |
+
"learning_rate": 3.3490788533847253e-05,
|
| 1037 |
+
"loss": 0.941,
|
| 1038 |
+
"step": 73500
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.997291141628819,
|
| 1042 |
+
"grad_norm": 2.6876494884490967,
|
| 1043 |
+
"learning_rate": 3.3378480972853015e-05,
|
| 1044 |
+
"loss": 0.9379,
|
| 1045 |
+
"step": 74000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 1.0040295952884732,
|
| 1049 |
+
"grad_norm": 2.9105379581451416,
|
| 1050 |
+
"learning_rate": 3.326617341185878e-05,
|
| 1051 |
+
"loss": 0.9376,
|
| 1052 |
+
"step": 74500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 1.0107680489481274,
|
| 1056 |
+
"grad_norm": 2.8576390743255615,
|
| 1057 |
+
"learning_rate": 3.3153865850864544e-05,
|
| 1058 |
+
"loss": 0.9368,
|
| 1059 |
+
"step": 75000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.0175065026077816,
|
| 1063 |
+
"grad_norm": 2.684570074081421,
|
| 1064 |
+
"learning_rate": 3.304155828987031e-05,
|
| 1065 |
+
"loss": 0.9365,
|
| 1066 |
+
"step": 75500
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.0242449562674358,
|
| 1070 |
+
"grad_norm": 2.759498119354248,
|
| 1071 |
+
"learning_rate": 3.292925072887607e-05,
|
| 1072 |
+
"loss": 0.9303,
|
| 1073 |
+
"step": 76000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 1.03098340992709,
|
| 1077 |
+
"grad_norm": 2.4174861907958984,
|
| 1078 |
+
"learning_rate": 3.2816943167881834e-05,
|
| 1079 |
+
"loss": 0.9342,
|
| 1080 |
+
"step": 76500
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 1.037721863586744,
|
| 1084 |
+
"grad_norm": 2.70633864402771,
|
| 1085 |
+
"learning_rate": 3.27046356068876e-05,
|
| 1086 |
+
"loss": 0.9277,
|
| 1087 |
+
"step": 77000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 1.0444603172463982,
|
| 1091 |
+
"grad_norm": 2.5511019229888916,
|
| 1092 |
+
"learning_rate": 3.259232804589336e-05,
|
| 1093 |
+
"loss": 0.9322,
|
| 1094 |
+
"step": 77500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.0511987709060524,
|
| 1098 |
+
"grad_norm": 2.8917860984802246,
|
| 1099 |
+
"learning_rate": 3.2480020484899124e-05,
|
| 1100 |
+
"loss": 0.9259,
|
| 1101 |
+
"step": 78000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.0579372245657066,
|
| 1105 |
+
"grad_norm": 2.6264562606811523,
|
| 1106 |
+
"learning_rate": 3.236771292390489e-05,
|
| 1107 |
+
"loss": 0.9249,
|
| 1108 |
+
"step": 78500
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 1.0646756782253608,
|
| 1112 |
+
"grad_norm": 2.544564962387085,
|
| 1113 |
+
"learning_rate": 3.225540536291066e-05,
|
| 1114 |
+
"loss": 0.9263,
|
| 1115 |
+
"step": 79000
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.071414131885015,
|
| 1119 |
+
"grad_norm": 2.7679262161254883,
|
| 1120 |
+
"learning_rate": 3.2143097801916414e-05,
|
| 1121 |
+
"loss": 0.9272,
|
| 1122 |
+
"step": 79500
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.0781525855446692,
|
| 1126 |
+
"grad_norm": 2.779067039489746,
|
| 1127 |
+
"learning_rate": 3.203079024092218e-05,
|
| 1128 |
+
"loss": 0.9228,
|
| 1129 |
+
"step": 80000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.0848910392043234,
|
| 1133 |
+
"grad_norm": 2.7541463375091553,
|
| 1134 |
+
"learning_rate": 3.191848267992795e-05,
|
| 1135 |
+
"loss": 0.9256,
|
| 1136 |
+
"step": 80500
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.0916294928639776,
|
| 1140 |
+
"grad_norm": 2.6884372234344482,
|
| 1141 |
+
"learning_rate": 3.1806175118933704e-05,
|
| 1142 |
+
"loss": 0.9238,
|
| 1143 |
+
"step": 81000
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.0983679465236318,
|
| 1147 |
+
"grad_norm": 2.8054563999176025,
|
| 1148 |
+
"learning_rate": 3.169386755793947e-05,
|
| 1149 |
+
"loss": 0.9164,
|
| 1150 |
+
"step": 81500
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 1.105106400183286,
|
| 1154 |
+
"grad_norm": 2.790738344192505,
|
| 1155 |
+
"learning_rate": 3.158155999694524e-05,
|
| 1156 |
+
"loss": 0.9139,
|
| 1157 |
+
"step": 82000
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.1118448538429402,
|
| 1161 |
+
"grad_norm": 2.869054079055786,
|
| 1162 |
+
"learning_rate": 3.1469252435951e-05,
|
| 1163 |
+
"loss": 0.92,
|
| 1164 |
+
"step": 82500
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.1185833075025944,
|
| 1168 |
+
"grad_norm": 2.479454517364502,
|
| 1169 |
+
"learning_rate": 3.135694487495676e-05,
|
| 1170 |
+
"loss": 0.9157,
|
| 1171 |
+
"step": 83000
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.1253217611622486,
|
| 1175 |
+
"grad_norm": 2.7569477558135986,
|
| 1176 |
+
"learning_rate": 3.124463731396253e-05,
|
| 1177 |
+
"loss": 0.9164,
|
| 1178 |
+
"step": 83500
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.1320602148219026,
|
| 1182 |
+
"grad_norm": 2.881066083908081,
|
| 1183 |
+
"learning_rate": 3.113232975296829e-05,
|
| 1184 |
+
"loss": 0.9185,
|
| 1185 |
+
"step": 84000
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.1387986684815568,
|
| 1189 |
+
"grad_norm": 2.7234609127044678,
|
| 1190 |
+
"learning_rate": 3.102002219197405e-05,
|
| 1191 |
+
"loss": 0.911,
|
| 1192 |
+
"step": 84500
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 1.145537122141211,
|
| 1196 |
+
"grad_norm": 2.785536766052246,
|
| 1197 |
+
"learning_rate": 3.090771463097982e-05,
|
| 1198 |
+
"loss": 0.9159,
|
| 1199 |
+
"step": 85000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.1522755758008651,
|
| 1203 |
+
"grad_norm": 3.004667282104492,
|
| 1204 |
+
"learning_rate": 3.079540706998558e-05,
|
| 1205 |
+
"loss": 0.9151,
|
| 1206 |
+
"step": 85500
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.1590140294605193,
|
| 1210 |
+
"grad_norm": 2.7210521697998047,
|
| 1211 |
+
"learning_rate": 3.068309950899135e-05,
|
| 1212 |
+
"loss": 0.9125,
|
| 1213 |
+
"step": 86000
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.1657524831201735,
|
| 1217 |
+
"grad_norm": 2.696476936340332,
|
| 1218 |
+
"learning_rate": 3.057079194799711e-05,
|
| 1219 |
+
"loss": 0.9122,
|
| 1220 |
+
"step": 86500
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 1.1724909367798277,
|
| 1224 |
+
"grad_norm": 2.855682849884033,
|
| 1225 |
+
"learning_rate": 3.045848438700287e-05,
|
| 1226 |
+
"loss": 0.9072,
|
| 1227 |
+
"step": 87000
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.179229390439482,
|
| 1231 |
+
"grad_norm": 2.6870203018188477,
|
| 1232 |
+
"learning_rate": 3.0346176826008636e-05,
|
| 1233 |
+
"loss": 0.9089,
|
| 1234 |
+
"step": 87500
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.1859678440991361,
|
| 1238 |
+
"grad_norm": 2.7807328701019287,
|
| 1239 |
+
"learning_rate": 3.0233869265014397e-05,
|
| 1240 |
+
"loss": 0.9043,
|
| 1241 |
+
"step": 88000
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.1927062977587903,
|
| 1245 |
+
"grad_norm": 2.4744486808776855,
|
| 1246 |
+
"learning_rate": 3.012156170402016e-05,
|
| 1247 |
+
"loss": 0.9004,
|
| 1248 |
+
"step": 88500
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.1994447514184445,
|
| 1252 |
+
"grad_norm": 2.7316670417785645,
|
| 1253 |
+
"learning_rate": 3.0009254143025926e-05,
|
| 1254 |
+
"loss": 0.9045,
|
| 1255 |
+
"step": 89000
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.2061832050780987,
|
| 1259 |
+
"grad_norm": 2.6756207942962646,
|
| 1260 |
+
"learning_rate": 2.9896946582031687e-05,
|
| 1261 |
+
"loss": 0.9062,
|
| 1262 |
+
"step": 89500
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.212921658737753,
|
| 1266 |
+
"grad_norm": 3.016295909881592,
|
| 1267 |
+
"learning_rate": 2.9784639021037452e-05,
|
| 1268 |
+
"loss": 0.9045,
|
| 1269 |
+
"step": 90000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.2196601123974071,
|
| 1273 |
+
"grad_norm": 2.6062419414520264,
|
| 1274 |
+
"learning_rate": 2.9672331460043216e-05,
|
| 1275 |
+
"loss": 0.905,
|
| 1276 |
+
"step": 90500
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 1.2263985660570613,
|
| 1280 |
+
"grad_norm": 2.849276065826416,
|
| 1281 |
+
"learning_rate": 2.9560023899048984e-05,
|
| 1282 |
+
"loss": 0.9048,
|
| 1283 |
+
"step": 91000
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 1.2331370197167155,
|
| 1287 |
+
"grad_norm": 3.0506439208984375,
|
| 1288 |
+
"learning_rate": 2.9447716338054742e-05,
|
| 1289 |
+
"loss": 0.8995,
|
| 1290 |
+
"step": 91500
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.2398754733763695,
|
| 1294 |
+
"grad_norm": 2.512368679046631,
|
| 1295 |
+
"learning_rate": 2.9335408777060506e-05,
|
| 1296 |
+
"loss": 0.8957,
|
| 1297 |
+
"step": 92000
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.2466139270360237,
|
| 1301 |
+
"grad_norm": 2.5880439281463623,
|
| 1302 |
+
"learning_rate": 2.9223101216066274e-05,
|
| 1303 |
+
"loss": 0.9048,
|
| 1304 |
+
"step": 92500
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 1.253352380695678,
|
| 1308 |
+
"grad_norm": 2.7087182998657227,
|
| 1309 |
+
"learning_rate": 2.9110793655072032e-05,
|
| 1310 |
+
"loss": 0.8984,
|
| 1311 |
+
"step": 93000
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.260090834355332,
|
| 1315 |
+
"grad_norm": 2.4637937545776367,
|
| 1316 |
+
"learning_rate": 2.8998486094077797e-05,
|
| 1317 |
+
"loss": 0.8951,
|
| 1318 |
+
"step": 93500
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 1.2668292880149863,
|
| 1322 |
+
"grad_norm": 2.6700010299682617,
|
| 1323 |
+
"learning_rate": 2.8886178533083564e-05,
|
| 1324 |
+
"loss": 0.8897,
|
| 1325 |
+
"step": 94000
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 1.2735677416746405,
|
| 1329 |
+
"grad_norm": 2.741194248199463,
|
| 1330 |
+
"learning_rate": 2.877387097208933e-05,
|
| 1331 |
+
"loss": 0.8914,
|
| 1332 |
+
"step": 94500
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 1.2803061953342947,
|
| 1336 |
+
"grad_norm": 2.547344446182251,
|
| 1337 |
+
"learning_rate": 2.8661563411095087e-05,
|
| 1338 |
+
"loss": 0.8868,
|
| 1339 |
+
"step": 95000
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 1.2870446489939489,
|
| 1343 |
+
"grad_norm": 2.5341854095458984,
|
| 1344 |
+
"learning_rate": 2.8549255850100855e-05,
|
| 1345 |
+
"loss": 0.888,
|
| 1346 |
+
"step": 95500
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 1.293783102653603,
|
| 1350 |
+
"grad_norm": 2.7855186462402344,
|
| 1351 |
+
"learning_rate": 2.843694828910662e-05,
|
| 1352 |
+
"loss": 0.8916,
|
| 1353 |
+
"step": 96000
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.3005215563132573,
|
| 1357 |
+
"grad_norm": 2.5844016075134277,
|
| 1358 |
+
"learning_rate": 2.8324640728112377e-05,
|
| 1359 |
+
"loss": 0.8903,
|
| 1360 |
+
"step": 96500
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 1.3072600099729115,
|
| 1364 |
+
"grad_norm": 2.713486671447754,
|
| 1365 |
+
"learning_rate": 2.8212333167118145e-05,
|
| 1366 |
+
"loss": 0.8872,
|
| 1367 |
+
"step": 97000
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 1.3139984636325657,
|
| 1371 |
+
"grad_norm": 2.8537466526031494,
|
| 1372 |
+
"learning_rate": 2.810002560612391e-05,
|
| 1373 |
+
"loss": 0.8856,
|
| 1374 |
+
"step": 97500
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.3207369172922196,
|
| 1378 |
+
"grad_norm": 3.3214476108551025,
|
| 1379 |
+
"learning_rate": 2.7987718045129674e-05,
|
| 1380 |
+
"loss": 0.8819,
|
| 1381 |
+
"step": 98000
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 1.3274753709518738,
|
| 1385 |
+
"grad_norm": 4.191791534423828,
|
| 1386 |
+
"learning_rate": 2.7875410484135435e-05,
|
| 1387 |
+
"loss": 0.885,
|
| 1388 |
+
"step": 98500
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 1.334213824611528,
|
| 1392 |
+
"grad_norm": 2.6634578704833984,
|
| 1393 |
+
"learning_rate": 2.77631029231412e-05,
|
| 1394 |
+
"loss": 0.8818,
|
| 1395 |
+
"step": 99000
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.3409522782711822,
|
| 1399 |
+
"grad_norm": 2.5315937995910645,
|
| 1400 |
+
"learning_rate": 2.7650795362146964e-05,
|
| 1401 |
+
"loss": 0.8883,
|
| 1402 |
+
"step": 99500
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 1.3476907319308364,
|
| 1406 |
+
"grad_norm": 2.7929606437683105,
|
| 1407 |
+
"learning_rate": 2.7538487801152725e-05,
|
| 1408 |
+
"loss": 0.8803,
|
| 1409 |
+
"step": 100000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 1.3544291855904906,
|
| 1413 |
+
"grad_norm": 2.932039499282837,
|
| 1414 |
+
"learning_rate": 2.742618024015849e-05,
|
| 1415 |
+
"loss": 0.8814,
|
| 1416 |
+
"step": 100500
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 1.3611676392501448,
|
| 1420 |
+
"grad_norm": 2.803535223007202,
|
| 1421 |
+
"learning_rate": 2.7313872679164254e-05,
|
| 1422 |
+
"loss": 0.8827,
|
| 1423 |
+
"step": 101000
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 1.367906092909799,
|
| 1427 |
+
"grad_norm": 2.8831472396850586,
|
| 1428 |
+
"learning_rate": 2.720156511817002e-05,
|
| 1429 |
+
"loss": 0.8813,
|
| 1430 |
+
"step": 101500
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 1.3746445465694532,
|
| 1434 |
+
"grad_norm": 2.710294723510742,
|
| 1435 |
+
"learning_rate": 2.708925755717578e-05,
|
| 1436 |
+
"loss": 0.8794,
|
| 1437 |
+
"step": 102000
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 1.3813830002291074,
|
| 1441 |
+
"grad_norm": 2.6942670345306396,
|
| 1442 |
+
"learning_rate": 2.6976949996181544e-05,
|
| 1443 |
+
"loss": 0.8782,
|
| 1444 |
+
"step": 102500
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 1.3881214538887616,
|
| 1448 |
+
"grad_norm": 2.7076780796051025,
|
| 1449 |
+
"learning_rate": 2.686464243518731e-05,
|
| 1450 |
+
"loss": 0.8778,
|
| 1451 |
+
"step": 103000
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 1.3948599075484158,
|
| 1455 |
+
"grad_norm": 2.730288028717041,
|
| 1456 |
+
"learning_rate": 2.675233487419307e-05,
|
| 1457 |
+
"loss": 0.8786,
|
| 1458 |
+
"step": 103500
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 1.40159836120807,
|
| 1462 |
+
"grad_norm": 2.5810985565185547,
|
| 1463 |
+
"learning_rate": 2.6640027313198834e-05,
|
| 1464 |
+
"loss": 0.8734,
|
| 1465 |
+
"step": 104000
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 1.4083368148677242,
|
| 1469 |
+
"grad_norm": 2.7088325023651123,
|
| 1470 |
+
"learning_rate": 2.65277197522046e-05,
|
| 1471 |
+
"loss": 0.8762,
|
| 1472 |
+
"step": 104500
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 1.4150752685273784,
|
| 1476 |
+
"grad_norm": 2.602531671524048,
|
| 1477 |
+
"learning_rate": 2.6415412191210363e-05,
|
| 1478 |
+
"loss": 0.8727,
|
| 1479 |
+
"step": 105000
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 1.4218137221870326,
|
| 1483 |
+
"grad_norm": 2.8586950302124023,
|
| 1484 |
+
"learning_rate": 2.6303104630216125e-05,
|
| 1485 |
+
"loss": 0.8769,
|
| 1486 |
+
"step": 105500
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 1.4285521758466868,
|
| 1490 |
+
"grad_norm": 2.7406837940216064,
|
| 1491 |
+
"learning_rate": 2.619079706922189e-05,
|
| 1492 |
+
"loss": 0.8707,
|
| 1493 |
+
"step": 106000
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 1.435290629506341,
|
| 1497 |
+
"grad_norm": 2.6816864013671875,
|
| 1498 |
+
"learning_rate": 2.6078489508227654e-05,
|
| 1499 |
+
"loss": 0.874,
|
| 1500 |
+
"step": 106500
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 1.442029083165995,
|
| 1504 |
+
"grad_norm": 2.88576078414917,
|
| 1505 |
+
"learning_rate": 2.5966181947233415e-05,
|
| 1506 |
+
"loss": 0.8679,
|
| 1507 |
+
"step": 107000
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 1.4487675368256492,
|
| 1511 |
+
"grad_norm": 2.5160343647003174,
|
| 1512 |
+
"learning_rate": 2.585387438623918e-05,
|
| 1513 |
+
"loss": 0.8687,
|
| 1514 |
+
"step": 107500
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 1.4555059904853034,
|
| 1518 |
+
"grad_norm": 2.585660934448242,
|
| 1519 |
+
"learning_rate": 2.5741566825244944e-05,
|
| 1520 |
+
"loss": 0.8659,
|
| 1521 |
+
"step": 108000
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 1.4622444441449576,
|
| 1525 |
+
"grad_norm": 2.697781801223755,
|
| 1526 |
+
"learning_rate": 2.562925926425071e-05,
|
| 1527 |
+
"loss": 0.8708,
|
| 1528 |
+
"step": 108500
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 1.4689828978046118,
|
| 1532 |
+
"grad_norm": 2.7062997817993164,
|
| 1533 |
+
"learning_rate": 2.551695170325647e-05,
|
| 1534 |
+
"loss": 0.867,
|
| 1535 |
+
"step": 109000
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 1.475721351464266,
|
| 1539 |
+
"grad_norm": 2.6003782749176025,
|
| 1540 |
+
"learning_rate": 2.5404644142262234e-05,
|
| 1541 |
+
"loss": 0.8747,
|
| 1542 |
+
"step": 109500
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 1.4824598051239202,
|
| 1546 |
+
"grad_norm": 3.052008628845215,
|
| 1547 |
+
"learning_rate": 2.5292336581268002e-05,
|
| 1548 |
+
"loss": 0.8667,
|
| 1549 |
+
"step": 110000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 1.4891982587835744,
|
| 1553 |
+
"grad_norm": 2.4037067890167236,
|
| 1554 |
+
"learning_rate": 2.518002902027376e-05,
|
| 1555 |
+
"loss": 0.8624,
|
| 1556 |
+
"step": 110500
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 1.4959367124432286,
|
| 1560 |
+
"grad_norm": 2.7442452907562256,
|
| 1561 |
+
"learning_rate": 2.5067721459279524e-05,
|
| 1562 |
+
"loss": 0.8674,
|
| 1563 |
+
"step": 111000
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 1.5026751661028828,
|
| 1567 |
+
"grad_norm": 2.840303659439087,
|
| 1568 |
+
"learning_rate": 2.4955413898285292e-05,
|
| 1569 |
+
"loss": 0.866,
|
| 1570 |
+
"step": 111500
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 1.5094136197625367,
|
| 1574 |
+
"grad_norm": 2.9296963214874268,
|
| 1575 |
+
"learning_rate": 2.4843106337291053e-05,
|
| 1576 |
+
"loss": 0.8641,
|
| 1577 |
+
"step": 112000
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 1.516152073422191,
|
| 1581 |
+
"grad_norm": 2.299018383026123,
|
| 1582 |
+
"learning_rate": 2.4730798776296818e-05,
|
| 1583 |
+
"loss": 0.8614,
|
| 1584 |
+
"step": 112500
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 1.5228905270818451,
|
| 1588 |
+
"grad_norm": 2.8484160900115967,
|
| 1589 |
+
"learning_rate": 2.4618491215302582e-05,
|
| 1590 |
+
"loss": 0.8656,
|
| 1591 |
+
"step": 113000
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 1.5296289807414993,
|
| 1595 |
+
"grad_norm": 2.9366495609283447,
|
| 1596 |
+
"learning_rate": 2.4506183654308343e-05,
|
| 1597 |
+
"loss": 0.8603,
|
| 1598 |
+
"step": 113500
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 1.5363674344011535,
|
| 1602 |
+
"grad_norm": 2.8661820888519287,
|
| 1603 |
+
"learning_rate": 2.4393876093314108e-05,
|
| 1604 |
+
"loss": 0.861,
|
| 1605 |
+
"step": 114000
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 1.5431058880608077,
|
| 1609 |
+
"grad_norm": 3.0576467514038086,
|
| 1610 |
+
"learning_rate": 2.4281568532319872e-05,
|
| 1611 |
+
"loss": 0.862,
|
| 1612 |
+
"step": 114500
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 1.549844341720462,
|
| 1616 |
+
"grad_norm": 2.5783610343933105,
|
| 1617 |
+
"learning_rate": 2.4169260971325637e-05,
|
| 1618 |
+
"loss": 0.856,
|
| 1619 |
+
"step": 115000
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 1.5565827953801161,
|
| 1623 |
+
"grad_norm": 2.6934993267059326,
|
| 1624 |
+
"learning_rate": 2.4056953410331398e-05,
|
| 1625 |
+
"loss": 0.8541,
|
| 1626 |
+
"step": 115500
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 1.5633212490397703,
|
| 1630 |
+
"grad_norm": 2.488381862640381,
|
| 1631 |
+
"learning_rate": 2.394464584933716e-05,
|
| 1632 |
+
"loss": 0.8615,
|
| 1633 |
+
"step": 116000
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 1.5700597026994245,
|
| 1637 |
+
"grad_norm": 2.6905736923217773,
|
| 1638 |
+
"learning_rate": 2.3832338288342927e-05,
|
| 1639 |
+
"loss": 0.8567,
|
| 1640 |
+
"step": 116500
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 1.5767981563590787,
|
| 1644 |
+
"grad_norm": 2.5146076679229736,
|
| 1645 |
+
"learning_rate": 2.3720030727348688e-05,
|
| 1646 |
+
"loss": 0.8585,
|
| 1647 |
+
"step": 117000
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 1.583536610018733,
|
| 1651 |
+
"grad_norm": 2.819962501525879,
|
| 1652 |
+
"learning_rate": 2.3607723166354452e-05,
|
| 1653 |
+
"loss": 0.8576,
|
| 1654 |
+
"step": 117500
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 1.5902750636783871,
|
| 1658 |
+
"grad_norm": 2.612535238265991,
|
| 1659 |
+
"learning_rate": 2.3495415605360217e-05,
|
| 1660 |
+
"loss": 0.8595,
|
| 1661 |
+
"step": 118000
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 1.5970135173380413,
|
| 1665 |
+
"grad_norm": 2.5329699516296387,
|
| 1666 |
+
"learning_rate": 2.338310804436598e-05,
|
| 1667 |
+
"loss": 0.857,
|
| 1668 |
+
"step": 118500
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 1.6037519709976955,
|
| 1672 |
+
"grad_norm": 3.002338409423828,
|
| 1673 |
+
"learning_rate": 2.3270800483371743e-05,
|
| 1674 |
+
"loss": 0.8503,
|
| 1675 |
+
"step": 119000
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 1.6104904246573497,
|
| 1679 |
+
"grad_norm": 2.588996648788452,
|
| 1680 |
+
"learning_rate": 2.3158492922377507e-05,
|
| 1681 |
+
"loss": 0.8565,
|
| 1682 |
+
"step": 119500
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 1.617228878317004,
|
| 1686 |
+
"grad_norm": 2.7498881816864014,
|
| 1687 |
+
"learning_rate": 2.304618536138327e-05,
|
| 1688 |
+
"loss": 0.8524,
|
| 1689 |
+
"step": 120000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 1.623967331976658,
|
| 1693 |
+
"grad_norm": 2.4528512954711914,
|
| 1694 |
+
"learning_rate": 2.2933877800389033e-05,
|
| 1695 |
+
"loss": 0.85,
|
| 1696 |
+
"step": 120500
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 1.6307057856363123,
|
| 1700 |
+
"grad_norm": 2.914182424545288,
|
| 1701 |
+
"learning_rate": 2.2821570239394797e-05,
|
| 1702 |
+
"loss": 0.8476,
|
| 1703 |
+
"step": 121000
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 1.6374442392959665,
|
| 1707 |
+
"grad_norm": 2.303053140640259,
|
| 1708 |
+
"learning_rate": 2.2709262678400562e-05,
|
| 1709 |
+
"loss": 0.851,
|
| 1710 |
+
"step": 121500
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 1.6441826929556207,
|
| 1714 |
+
"grad_norm": 2.736271381378174,
|
| 1715 |
+
"learning_rate": 2.2596955117406326e-05,
|
| 1716 |
+
"loss": 0.8496,
|
| 1717 |
+
"step": 122000
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 1.650921146615275,
|
| 1721 |
+
"grad_norm": 3.0336098670959473,
|
| 1722 |
+
"learning_rate": 2.2484647556412087e-05,
|
| 1723 |
+
"loss": 0.8449,
|
| 1724 |
+
"step": 122500
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 1.6576596002749289,
|
| 1728 |
+
"grad_norm": 3.153146505355835,
|
| 1729 |
+
"learning_rate": 2.2372339995417852e-05,
|
| 1730 |
+
"loss": 0.8484,
|
| 1731 |
+
"step": 123000
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 1.664398053934583,
|
| 1735 |
+
"grad_norm": 2.434753179550171,
|
| 1736 |
+
"learning_rate": 2.2260032434423616e-05,
|
| 1737 |
+
"loss": 0.8477,
|
| 1738 |
+
"step": 123500
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 1.6711365075942373,
|
| 1742 |
+
"grad_norm": 2.798680543899536,
|
| 1743 |
+
"learning_rate": 2.2147724873429378e-05,
|
| 1744 |
+
"loss": 0.8459,
|
| 1745 |
+
"step": 124000
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 1.6778749612538915,
|
| 1749 |
+
"grad_norm": 2.9313242435455322,
|
| 1750 |
+
"learning_rate": 2.2035417312435145e-05,
|
| 1751 |
+
"loss": 0.8484,
|
| 1752 |
+
"step": 124500
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 1.6846134149135457,
|
| 1756 |
+
"grad_norm": 2.831214189529419,
|
| 1757 |
+
"learning_rate": 2.1923109751440907e-05,
|
| 1758 |
+
"loss": 0.8438,
|
| 1759 |
+
"step": 125000
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 1.6913518685731999,
|
| 1763 |
+
"grad_norm": 3.033341407775879,
|
| 1764 |
+
"learning_rate": 2.181080219044667e-05,
|
| 1765 |
+
"loss": 0.8427,
|
| 1766 |
+
"step": 125500
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 1.698090322232854,
|
| 1770 |
+
"grad_norm": 2.704768657684326,
|
| 1771 |
+
"learning_rate": 2.1698494629452436e-05,
|
| 1772 |
+
"loss": 0.8429,
|
| 1773 |
+
"step": 126000
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 1.704828775892508,
|
| 1777 |
+
"grad_norm": 2.62070894241333,
|
| 1778 |
+
"learning_rate": 2.1586187068458197e-05,
|
| 1779 |
+
"loss": 0.8446,
|
| 1780 |
+
"step": 126500
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 1.7115672295521622,
|
| 1784 |
+
"grad_norm": 3.01562237739563,
|
| 1785 |
+
"learning_rate": 2.147387950746396e-05,
|
| 1786 |
+
"loss": 0.8433,
|
| 1787 |
+
"step": 127000
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 1.7183056832118164,
|
| 1791 |
+
"grad_norm": 2.577298641204834,
|
| 1792 |
+
"learning_rate": 2.1361571946469726e-05,
|
| 1793 |
+
"loss": 0.8444,
|
| 1794 |
+
"step": 127500
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 1.7250441368714706,
|
| 1798 |
+
"grad_norm": 2.642275810241699,
|
| 1799 |
+
"learning_rate": 2.124926438547549e-05,
|
| 1800 |
+
"loss": 0.8407,
|
| 1801 |
+
"step": 128000
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 1.7317825905311248,
|
| 1805 |
+
"grad_norm": 2.7020041942596436,
|
| 1806 |
+
"learning_rate": 2.113695682448125e-05,
|
| 1807 |
+
"loss": 0.842,
|
| 1808 |
+
"step": 128500
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 1.738521044190779,
|
| 1812 |
+
"grad_norm": 2.7257561683654785,
|
| 1813 |
+
"learning_rate": 2.1024649263487016e-05,
|
| 1814 |
+
"loss": 0.8444,
|
| 1815 |
+
"step": 129000
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 1.7452594978504332,
|
| 1819 |
+
"grad_norm": 2.3752248287200928,
|
| 1820 |
+
"learning_rate": 2.091234170249278e-05,
|
| 1821 |
+
"loss": 0.8387,
|
| 1822 |
+
"step": 129500
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 1.7519979515100874,
|
| 1826 |
+
"grad_norm": 2.5963971614837646,
|
| 1827 |
+
"learning_rate": 2.080003414149854e-05,
|
| 1828 |
+
"loss": 0.8381,
|
| 1829 |
+
"step": 130000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 1.7587364051697416,
|
| 1833 |
+
"grad_norm": 2.735293388366699,
|
| 1834 |
+
"learning_rate": 2.0687726580504306e-05,
|
| 1835 |
+
"loss": 0.8358,
|
| 1836 |
+
"step": 130500
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 1.7654748588293958,
|
| 1840 |
+
"grad_norm": 2.578404188156128,
|
| 1841 |
+
"learning_rate": 2.057541901951007e-05,
|
| 1842 |
+
"loss": 0.8433,
|
| 1843 |
+
"step": 131000
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 1.77221331248905,
|
| 1847 |
+
"grad_norm": 2.714272975921631,
|
| 1848 |
+
"learning_rate": 2.0463111458515835e-05,
|
| 1849 |
+
"loss": 0.8375,
|
| 1850 |
+
"step": 131500
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 1.7789517661487042,
|
| 1854 |
+
"grad_norm": 2.9564433097839355,
|
| 1855 |
+
"learning_rate": 2.0350803897521596e-05,
|
| 1856 |
+
"loss": 0.8387,
|
| 1857 |
+
"step": 132000
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 1.7856902198083584,
|
| 1861 |
+
"grad_norm": 2.411444902420044,
|
| 1862 |
+
"learning_rate": 2.023849633652736e-05,
|
| 1863 |
+
"loss": 0.8343,
|
| 1864 |
+
"step": 132500
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 1.7924286734680126,
|
| 1868 |
+
"grad_norm": 2.7173373699188232,
|
| 1869 |
+
"learning_rate": 2.0126188775533125e-05,
|
| 1870 |
+
"loss": 0.835,
|
| 1871 |
+
"step": 133000
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 1.7991671271276668,
|
| 1875 |
+
"grad_norm": 2.5802862644195557,
|
| 1876 |
+
"learning_rate": 2.0013881214538886e-05,
|
| 1877 |
+
"loss": 0.8326,
|
| 1878 |
+
"step": 133500
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 1.805905580787321,
|
| 1882 |
+
"grad_norm": 2.4736204147338867,
|
| 1883 |
+
"learning_rate": 1.9901573653544654e-05,
|
| 1884 |
+
"loss": 0.8339,
|
| 1885 |
+
"step": 134000
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 1.8126440344469752,
|
| 1889 |
+
"grad_norm": 2.9207983016967773,
|
| 1890 |
+
"learning_rate": 1.9789266092550415e-05,
|
| 1891 |
+
"loss": 0.8306,
|
| 1892 |
+
"step": 134500
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 1.8193824881066294,
|
| 1896 |
+
"grad_norm": 2.806415319442749,
|
| 1897 |
+
"learning_rate": 1.967695853155618e-05,
|
| 1898 |
+
"loss": 0.83,
|
| 1899 |
+
"step": 135000
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 1.8261209417662836,
|
| 1903 |
+
"grad_norm": 2.5810861587524414,
|
| 1904 |
+
"learning_rate": 1.9564650970561944e-05,
|
| 1905 |
+
"loss": 0.8318,
|
| 1906 |
+
"step": 135500
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 1.8328593954259378,
|
| 1910 |
+
"grad_norm": 2.784264087677002,
|
| 1911 |
+
"learning_rate": 1.9452343409567706e-05,
|
| 1912 |
+
"loss": 0.8343,
|
| 1913 |
+
"step": 136000
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 1.839597849085592,
|
| 1917 |
+
"grad_norm": 2.756681203842163,
|
| 1918 |
+
"learning_rate": 1.934003584857347e-05,
|
| 1919 |
+
"loss": 0.8282,
|
| 1920 |
+
"step": 136500
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 1.8463363027452462,
|
| 1924 |
+
"grad_norm": 3.2176573276519775,
|
| 1925 |
+
"learning_rate": 1.9227728287579235e-05,
|
| 1926 |
+
"loss": 0.8309,
|
| 1927 |
+
"step": 137000
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 1.8530747564049002,
|
| 1931 |
+
"grad_norm": 2.4733126163482666,
|
| 1932 |
+
"learning_rate": 1.9115420726585e-05,
|
| 1933 |
+
"loss": 0.8291,
|
| 1934 |
+
"step": 137500
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 1.8598132100645544,
|
| 1938 |
+
"grad_norm": 2.550459861755371,
|
| 1939 |
+
"learning_rate": 1.900311316559076e-05,
|
| 1940 |
+
"loss": 0.8328,
|
| 1941 |
+
"step": 138000
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 1.8665516637242086,
|
| 1945 |
+
"grad_norm": 2.9665493965148926,
|
| 1946 |
+
"learning_rate": 1.8890805604596525e-05,
|
| 1947 |
+
"loss": 0.8315,
|
| 1948 |
+
"step": 138500
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 1.8732901173838628,
|
| 1952 |
+
"grad_norm": 2.613665819168091,
|
| 1953 |
+
"learning_rate": 1.877849804360229e-05,
|
| 1954 |
+
"loss": 0.8287,
|
| 1955 |
+
"step": 139000
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 1.880028571043517,
|
| 1959 |
+
"grad_norm": 2.4834043979644775,
|
| 1960 |
+
"learning_rate": 1.866619048260805e-05,
|
| 1961 |
+
"loss": 0.828,
|
| 1962 |
+
"step": 139500
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 1.8867670247031711,
|
| 1966 |
+
"grad_norm": 2.3651483058929443,
|
| 1967 |
+
"learning_rate": 1.8553882921613815e-05,
|
| 1968 |
+
"loss": 0.8281,
|
| 1969 |
+
"step": 140000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 1.8935054783628253,
|
| 1973 |
+
"grad_norm": 2.588747501373291,
|
| 1974 |
+
"learning_rate": 1.844157536061958e-05,
|
| 1975 |
+
"loss": 0.827,
|
| 1976 |
+
"step": 140500
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 1.9002439320224793,
|
| 1980 |
+
"grad_norm": 2.5978434085845947,
|
| 1981 |
+
"learning_rate": 1.8329267799625344e-05,
|
| 1982 |
+
"loss": 0.8265,
|
| 1983 |
+
"step": 141000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 1.9069823856821335,
|
| 1987 |
+
"grad_norm": 2.6672565937042236,
|
| 1988 |
+
"learning_rate": 1.8216960238631105e-05,
|
| 1989 |
+
"loss": 0.8278,
|
| 1990 |
+
"step": 141500
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 1.9137208393417877,
|
| 1994 |
+
"grad_norm": 2.658770799636841,
|
| 1995 |
+
"learning_rate": 1.8104652677636873e-05,
|
| 1996 |
+
"loss": 0.8282,
|
| 1997 |
+
"step": 142000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 1.920459293001442,
|
| 2001 |
+
"grad_norm": 2.7485387325286865,
|
| 2002 |
+
"learning_rate": 1.7992345116642634e-05,
|
| 2003 |
+
"loss": 0.8251,
|
| 2004 |
+
"step": 142500
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 1.9271977466610961,
|
| 2008 |
+
"grad_norm": 2.95625376701355,
|
| 2009 |
+
"learning_rate": 1.7880037555648395e-05,
|
| 2010 |
+
"loss": 0.8208,
|
| 2011 |
+
"step": 143000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 1.9339362003207503,
|
| 2015 |
+
"grad_norm": 2.980334997177124,
|
| 2016 |
+
"learning_rate": 1.7767729994654163e-05,
|
| 2017 |
+
"loss": 0.8191,
|
| 2018 |
+
"step": 143500
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 1.9406746539804045,
|
| 2022 |
+
"grad_norm": 2.575261116027832,
|
| 2023 |
+
"learning_rate": 1.7655422433659924e-05,
|
| 2024 |
+
"loss": 0.8205,
|
| 2025 |
+
"step": 144000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 1.9474131076400587,
|
| 2029 |
+
"grad_norm": 3.0678629875183105,
|
| 2030 |
+
"learning_rate": 1.754311487266569e-05,
|
| 2031 |
+
"loss": 0.8186,
|
| 2032 |
+
"step": 144500
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 1.954151561299713,
|
| 2036 |
+
"grad_norm": 2.7946627140045166,
|
| 2037 |
+
"learning_rate": 1.7430807311671453e-05,
|
| 2038 |
+
"loss": 0.8203,
|
| 2039 |
+
"step": 145000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 1.960890014959367,
|
| 2043 |
+
"grad_norm": 2.817331075668335,
|
| 2044 |
+
"learning_rate": 1.7318499750677218e-05,
|
| 2045 |
+
"loss": 0.8186,
|
| 2046 |
+
"step": 145500
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 1.9676284686190213,
|
| 2050 |
+
"grad_norm": 2.7685699462890625,
|
| 2051 |
+
"learning_rate": 1.720619218968298e-05,
|
| 2052 |
+
"loss": 0.8217,
|
| 2053 |
+
"step": 146000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 1.9743669222786755,
|
| 2057 |
+
"grad_norm": 2.7003917694091797,
|
| 2058 |
+
"learning_rate": 1.7093884628688743e-05,
|
| 2059 |
+
"loss": 0.8206,
|
| 2060 |
+
"step": 146500
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 1.9811053759383297,
|
| 2064 |
+
"grad_norm": 2.629485845565796,
|
| 2065 |
+
"learning_rate": 1.6981577067694508e-05,
|
| 2066 |
+
"loss": 0.82,
|
| 2067 |
+
"step": 147000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 1.987843829597984,
|
| 2071 |
+
"grad_norm": 2.6622962951660156,
|
| 2072 |
+
"learning_rate": 1.686926950670027e-05,
|
| 2073 |
+
"loss": 0.8188,
|
| 2074 |
+
"step": 147500
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 1.994582283257638,
|
| 2078 |
+
"grad_norm": 2.8867058753967285,
|
| 2079 |
+
"learning_rate": 1.6756961945706033e-05,
|
| 2080 |
+
"loss": 0.819,
|
| 2081 |
+
"step": 148000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 2.0013207369172923,
|
| 2085 |
+
"grad_norm": 2.781660318374634,
|
| 2086 |
+
"learning_rate": 1.6644654384711798e-05,
|
| 2087 |
+
"loss": 0.8175,
|
| 2088 |
+
"step": 148500
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 2.0080591905769465,
|
| 2092 |
+
"grad_norm": 2.642421245574951,
|
| 2093 |
+
"learning_rate": 1.653234682371756e-05,
|
| 2094 |
+
"loss": 0.8131,
|
| 2095 |
+
"step": 149000
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 2.0147976442366007,
|
| 2099 |
+
"grad_norm": 2.9073877334594727,
|
| 2100 |
+
"learning_rate": 1.6420039262723324e-05,
|
| 2101 |
+
"loss": 0.816,
|
| 2102 |
+
"step": 149500
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 2.021536097896255,
|
| 2106 |
+
"grad_norm": 2.7006802558898926,
|
| 2107 |
+
"learning_rate": 1.6307731701729088e-05,
|
| 2108 |
+
"loss": 0.8152,
|
| 2109 |
+
"step": 150000
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 2.028274551555909,
|
| 2113 |
+
"grad_norm": 2.617917776107788,
|
| 2114 |
+
"learning_rate": 1.6195424140734853e-05,
|
| 2115 |
+
"loss": 0.8188,
|
| 2116 |
+
"step": 150500
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 2.0350130052155633,
|
| 2120 |
+
"grad_norm": 2.5655267238616943,
|
| 2121 |
+
"learning_rate": 1.6083116579740614e-05,
|
| 2122 |
+
"loss": 0.8117,
|
| 2123 |
+
"step": 151000
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 2.0417514588752175,
|
| 2127 |
+
"grad_norm": 2.7315967082977295,
|
| 2128 |
+
"learning_rate": 1.5970809018746378e-05,
|
| 2129 |
+
"loss": 0.8116,
|
| 2130 |
+
"step": 151500
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 2.0484899125348717,
|
| 2134 |
+
"grad_norm": 2.7256932258605957,
|
| 2135 |
+
"learning_rate": 1.5858501457752143e-05,
|
| 2136 |
+
"loss": 0.8081,
|
| 2137 |
+
"step": 152000
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 2.055228366194526,
|
| 2141 |
+
"grad_norm": 2.810154676437378,
|
| 2142 |
+
"learning_rate": 1.5746193896757904e-05,
|
| 2143 |
+
"loss": 0.8106,
|
| 2144 |
+
"step": 152500
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 2.06196681985418,
|
| 2148 |
+
"grad_norm": 2.749325752258301,
|
| 2149 |
+
"learning_rate": 1.563388633576367e-05,
|
| 2150 |
+
"loss": 0.8174,
|
| 2151 |
+
"step": 153000
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 2.0687052735138343,
|
| 2155 |
+
"grad_norm": 2.8705825805664062,
|
| 2156 |
+
"learning_rate": 1.5521578774769433e-05,
|
| 2157 |
+
"loss": 0.8119,
|
| 2158 |
+
"step": 153500
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 2.075443727173488,
|
| 2162 |
+
"grad_norm": 2.669269323348999,
|
| 2163 |
+
"learning_rate": 1.5409271213775197e-05,
|
| 2164 |
+
"loss": 0.8145,
|
| 2165 |
+
"step": 154000
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 2.082182180833142,
|
| 2169 |
+
"grad_norm": 2.9345993995666504,
|
| 2170 |
+
"learning_rate": 1.529696365278096e-05,
|
| 2171 |
+
"loss": 0.8085,
|
| 2172 |
+
"step": 154500
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 2.0889206344927964,
|
| 2176 |
+
"grad_norm": 2.713247299194336,
|
| 2177 |
+
"learning_rate": 1.5184656091786725e-05,
|
| 2178 |
+
"loss": 0.8087,
|
| 2179 |
+
"step": 155000
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 2.0956590881524506,
|
| 2183 |
+
"grad_norm": 2.6601617336273193,
|
| 2184 |
+
"learning_rate": 1.5072348530792488e-05,
|
| 2185 |
+
"loss": 0.8015,
|
| 2186 |
+
"step": 155500
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 2.102397541812105,
|
| 2190 |
+
"grad_norm": 3.3940951824188232,
|
| 2191 |
+
"learning_rate": 1.496004096979825e-05,
|
| 2192 |
+
"loss": 0.8075,
|
| 2193 |
+
"step": 156000
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 2.109135995471759,
|
| 2197 |
+
"grad_norm": 2.4791245460510254,
|
| 2198 |
+
"learning_rate": 1.4847733408804015e-05,
|
| 2199 |
+
"loss": 0.8075,
|
| 2200 |
+
"step": 156500
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 2.115874449131413,
|
| 2204 |
+
"grad_norm": 3.14079213142395,
|
| 2205 |
+
"learning_rate": 1.4735425847809778e-05,
|
| 2206 |
+
"loss": 0.8051,
|
| 2207 |
+
"step": 157000
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 2.1226129027910674,
|
| 2211 |
+
"grad_norm": 2.716848611831665,
|
| 2212 |
+
"learning_rate": 1.4623118286815542e-05,
|
| 2213 |
+
"loss": 0.8068,
|
| 2214 |
+
"step": 157500
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 2.1293513564507216,
|
| 2218 |
+
"grad_norm": 2.691509246826172,
|
| 2219 |
+
"learning_rate": 1.4510810725821305e-05,
|
| 2220 |
+
"loss": 0.8081,
|
| 2221 |
+
"step": 158000
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 2.136089810110376,
|
| 2225 |
+
"grad_norm": 2.5966506004333496,
|
| 2226 |
+
"learning_rate": 1.4398503164827071e-05,
|
| 2227 |
+
"loss": 0.8071,
|
| 2228 |
+
"step": 158500
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 2.14282826377003,
|
| 2232 |
+
"grad_norm": 2.498093605041504,
|
| 2233 |
+
"learning_rate": 1.4286195603832832e-05,
|
| 2234 |
+
"loss": 0.8081,
|
| 2235 |
+
"step": 159000
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 2.149566717429684,
|
| 2239 |
+
"grad_norm": 3.0157721042633057,
|
| 2240 |
+
"learning_rate": 1.4173888042838595e-05,
|
| 2241 |
+
"loss": 0.8066,
|
| 2242 |
+
"step": 159500
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 2.1563051710893384,
|
| 2246 |
+
"grad_norm": 2.4180946350097656,
|
| 2247 |
+
"learning_rate": 1.4061580481844361e-05,
|
| 2248 |
+
"loss": 0.8042,
|
| 2249 |
+
"step": 160000
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 2.1630436247489926,
|
| 2253 |
+
"grad_norm": 2.5269594192504883,
|
| 2254 |
+
"learning_rate": 1.3949272920850123e-05,
|
| 2255 |
+
"loss": 0.8048,
|
| 2256 |
+
"step": 160500
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 2.169782078408647,
|
| 2260 |
+
"grad_norm": 2.8210670948028564,
|
| 2261 |
+
"learning_rate": 1.3836965359855889e-05,
|
| 2262 |
+
"loss": 0.8057,
|
| 2263 |
+
"step": 161000
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 2.176520532068301,
|
| 2267 |
+
"grad_norm": 2.71695876121521,
|
| 2268 |
+
"learning_rate": 1.3724657798861652e-05,
|
| 2269 |
+
"loss": 0.8027,
|
| 2270 |
+
"step": 161500
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 2.183258985727955,
|
| 2274 |
+
"grad_norm": 2.7251980304718018,
|
| 2275 |
+
"learning_rate": 1.3612350237867416e-05,
|
| 2276 |
+
"loss": 0.8058,
|
| 2277 |
+
"step": 162000
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 2.1899974393876094,
|
| 2281 |
+
"grad_norm": 2.8365695476531982,
|
| 2282 |
+
"learning_rate": 1.3500042676873179e-05,
|
| 2283 |
+
"loss": 0.8012,
|
| 2284 |
+
"step": 162500
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 2.1967358930472636,
|
| 2288 |
+
"grad_norm": 2.5428919792175293,
|
| 2289 |
+
"learning_rate": 1.3387735115878942e-05,
|
| 2290 |
+
"loss": 0.8028,
|
| 2291 |
+
"step": 163000
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 2.2034743467069178,
|
| 2295 |
+
"grad_norm": 2.581951379776001,
|
| 2296 |
+
"learning_rate": 1.3275427554884706e-05,
|
| 2297 |
+
"loss": 0.7992,
|
| 2298 |
+
"step": 163500
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 2.210212800366572,
|
| 2302 |
+
"grad_norm": 2.654918909072876,
|
| 2303 |
+
"learning_rate": 1.3163119993890469e-05,
|
| 2304 |
+
"loss": 0.8014,
|
| 2305 |
+
"step": 164000
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 2.216951254026226,
|
| 2309 |
+
"grad_norm": 2.552696704864502,
|
| 2310 |
+
"learning_rate": 1.3050812432896234e-05,
|
| 2311 |
+
"loss": 0.8066,
|
| 2312 |
+
"step": 164500
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 2.2236897076858804,
|
| 2316 |
+
"grad_norm": 2.363924980163574,
|
| 2317 |
+
"learning_rate": 1.2938504871901996e-05,
|
| 2318 |
+
"loss": 0.8002,
|
| 2319 |
+
"step": 165000
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 2.2304281613455346,
|
| 2323 |
+
"grad_norm": 2.6942362785339355,
|
| 2324 |
+
"learning_rate": 1.2826197310907761e-05,
|
| 2325 |
+
"loss": 0.798,
|
| 2326 |
+
"step": 165500
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 2.2371666150051888,
|
| 2330 |
+
"grad_norm": 2.5158755779266357,
|
| 2331 |
+
"learning_rate": 1.2713889749913524e-05,
|
| 2332 |
+
"loss": 0.8,
|
| 2333 |
+
"step": 166000
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 2.243905068664843,
|
| 2337 |
+
"grad_norm": 3.532792329788208,
|
| 2338 |
+
"learning_rate": 1.2601582188919286e-05,
|
| 2339 |
+
"loss": 0.7952,
|
| 2340 |
+
"step": 166500
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 2.250643522324497,
|
| 2344 |
+
"grad_norm": 2.8137147426605225,
|
| 2345 |
+
"learning_rate": 1.2489274627925051e-05,
|
| 2346 |
+
"loss": 0.8014,
|
| 2347 |
+
"step": 167000
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 2.257381975984151,
|
| 2351 |
+
"grad_norm": 3.056257486343384,
|
| 2352 |
+
"learning_rate": 1.2376967066930814e-05,
|
| 2353 |
+
"loss": 0.7989,
|
| 2354 |
+
"step": 167500
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 2.264120429643805,
|
| 2358 |
+
"grad_norm": 3.0764172077178955,
|
| 2359 |
+
"learning_rate": 1.2264659505936578e-05,
|
| 2360 |
+
"loss": 0.7963,
|
| 2361 |
+
"step": 168000
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 2.2708588833034593,
|
| 2365 |
+
"grad_norm": 3.017273187637329,
|
| 2366 |
+
"learning_rate": 1.2152351944942341e-05,
|
| 2367 |
+
"loss": 0.7968,
|
| 2368 |
+
"step": 168500
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 2.2775973369631135,
|
| 2372 |
+
"grad_norm": 2.690185308456421,
|
| 2373 |
+
"learning_rate": 1.2040044383948106e-05,
|
| 2374 |
+
"loss": 0.7968,
|
| 2375 |
+
"step": 169000
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 2.2843357906227677,
|
| 2379 |
+
"grad_norm": 2.7114124298095703,
|
| 2380 |
+
"learning_rate": 1.192773682295387e-05,
|
| 2381 |
+
"loss": 0.8011,
|
| 2382 |
+
"step": 169500
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 2.291074244282422,
|
| 2386 |
+
"grad_norm": 2.9144785404205322,
|
| 2387 |
+
"learning_rate": 1.1815429261959633e-05,
|
| 2388 |
+
"loss": 0.7963,
|
| 2389 |
+
"step": 170000
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 2.297812697942076,
|
| 2393 |
+
"grad_norm": 2.8431060314178467,
|
| 2394 |
+
"learning_rate": 1.1703121700965396e-05,
|
| 2395 |
+
"loss": 0.7927,
|
| 2396 |
+
"step": 170500
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 2.3045511516017303,
|
| 2400 |
+
"grad_norm": 2.4635190963745117,
|
| 2401 |
+
"learning_rate": 1.159081413997116e-05,
|
| 2402 |
+
"loss": 0.796,
|
| 2403 |
+
"step": 171000
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 2.3112896052613845,
|
| 2407 |
+
"grad_norm": 2.829127788543701,
|
| 2408 |
+
"learning_rate": 1.1478506578976923e-05,
|
| 2409 |
+
"loss": 0.7933,
|
| 2410 |
+
"step": 171500
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 2.3180280589210387,
|
| 2414 |
+
"grad_norm": 2.748392343521118,
|
| 2415 |
+
"learning_rate": 1.1366199017982688e-05,
|
| 2416 |
+
"loss": 0.7982,
|
| 2417 |
+
"step": 172000
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 2.324766512580693,
|
| 2421 |
+
"grad_norm": 2.6858222484588623,
|
| 2422 |
+
"learning_rate": 1.125389145698845e-05,
|
| 2423 |
+
"loss": 0.7919,
|
| 2424 |
+
"step": 172500
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 2.331504966240347,
|
| 2428 |
+
"grad_norm": 2.5816309452056885,
|
| 2429 |
+
"learning_rate": 1.1141583895994215e-05,
|
| 2430 |
+
"loss": 0.7943,
|
| 2431 |
+
"step": 173000
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 2.3382434199000013,
|
| 2435 |
+
"grad_norm": 2.663086175918579,
|
| 2436 |
+
"learning_rate": 1.1029276334999978e-05,
|
| 2437 |
+
"loss": 0.7907,
|
| 2438 |
+
"step": 173500
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 2.3449818735596555,
|
| 2442 |
+
"grad_norm": 2.7774062156677246,
|
| 2443 |
+
"learning_rate": 1.091696877400574e-05,
|
| 2444 |
+
"loss": 0.7938,
|
| 2445 |
+
"step": 174000
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 2.3517203272193097,
|
| 2449 |
+
"grad_norm": 2.579991102218628,
|
| 2450 |
+
"learning_rate": 1.0804661213011505e-05,
|
| 2451 |
+
"loss": 0.7907,
|
| 2452 |
+
"step": 174500
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 2.358458780878964,
|
| 2456 |
+
"grad_norm": 2.737271547317505,
|
| 2457 |
+
"learning_rate": 1.0692353652017268e-05,
|
| 2458 |
+
"loss": 0.791,
|
| 2459 |
+
"step": 175000
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 2.365197234538618,
|
| 2463 |
+
"grad_norm": 2.7510414123535156,
|
| 2464 |
+
"learning_rate": 1.0580046091023032e-05,
|
| 2465 |
+
"loss": 0.7897,
|
| 2466 |
+
"step": 175500
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 2.3719356881982723,
|
| 2470 |
+
"grad_norm": 2.9037694931030273,
|
| 2471 |
+
"learning_rate": 1.0467738530028797e-05,
|
| 2472 |
+
"loss": 0.7911,
|
| 2473 |
+
"step": 176000
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 2.3786741418579265,
|
| 2477 |
+
"grad_norm": 2.720240354537964,
|
| 2478 |
+
"learning_rate": 1.035543096903456e-05,
|
| 2479 |
+
"loss": 0.7927,
|
| 2480 |
+
"step": 176500
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 2.3854125955175807,
|
| 2484 |
+
"grad_norm": 2.5593862533569336,
|
| 2485 |
+
"learning_rate": 1.0243123408040324e-05,
|
| 2486 |
+
"loss": 0.7884,
|
| 2487 |
+
"step": 177000
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 2.392151049177235,
|
| 2491 |
+
"grad_norm": 2.8841631412506104,
|
| 2492 |
+
"learning_rate": 1.0130815847046087e-05,
|
| 2493 |
+
"loss": 0.7884,
|
| 2494 |
+
"step": 177500
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 2.398889502836889,
|
| 2498 |
+
"grad_norm": 2.46235728263855,
|
| 2499 |
+
"learning_rate": 1.001850828605185e-05,
|
| 2500 |
+
"loss": 0.7849,
|
| 2501 |
+
"step": 178000
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 2.4056279564965433,
|
| 2505 |
+
"grad_norm": 2.6925342082977295,
|
| 2506 |
+
"learning_rate": 9.906200725057614e-06,
|
| 2507 |
+
"loss": 0.787,
|
| 2508 |
+
"step": 178500
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 2.4123664101561975,
|
| 2512 |
+
"grad_norm": 2.830958366394043,
|
| 2513 |
+
"learning_rate": 9.793893164063377e-06,
|
| 2514 |
+
"loss": 0.7851,
|
| 2515 |
+
"step": 179000
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 2.4191048638158517,
|
| 2519 |
+
"grad_norm": 2.656013011932373,
|
| 2520 |
+
"learning_rate": 9.681585603069142e-06,
|
| 2521 |
+
"loss": 0.7869,
|
| 2522 |
+
"step": 179500
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 2.425843317475506,
|
| 2526 |
+
"grad_norm": 2.9063632488250732,
|
| 2527 |
+
"learning_rate": 9.569278042074906e-06,
|
| 2528 |
+
"loss": 0.7835,
|
| 2529 |
+
"step": 180000
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 2.43258177113516,
|
| 2533 |
+
"grad_norm": 2.727149724960327,
|
| 2534 |
+
"learning_rate": 9.456970481080667e-06,
|
| 2535 |
+
"loss": 0.7872,
|
| 2536 |
+
"step": 180500
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 2.4393202247948143,
|
| 2540 |
+
"grad_norm": 2.8738434314727783,
|
| 2541 |
+
"learning_rate": 9.344662920086432e-06,
|
| 2542 |
+
"loss": 0.7893,
|
| 2543 |
+
"step": 181000
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 2.4460586784544684,
|
| 2547 |
+
"grad_norm": 2.6790006160736084,
|
| 2548 |
+
"learning_rate": 9.232355359092196e-06,
|
| 2549 |
+
"loss": 0.7901,
|
| 2550 |
+
"step": 181500
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 2.4527971321141226,
|
| 2554 |
+
"grad_norm": 2.999220371246338,
|
| 2555 |
+
"learning_rate": 9.12004779809796e-06,
|
| 2556 |
+
"loss": 0.7846,
|
| 2557 |
+
"step": 182000
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 2.459535585773777,
|
| 2561 |
+
"grad_norm": 2.930610418319702,
|
| 2562 |
+
"learning_rate": 9.007740237103724e-06,
|
| 2563 |
+
"loss": 0.7836,
|
| 2564 |
+
"step": 182500
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 2.466274039433431,
|
| 2568 |
+
"grad_norm": 2.5325422286987305,
|
| 2569 |
+
"learning_rate": 8.895432676109487e-06,
|
| 2570 |
+
"loss": 0.7841,
|
| 2571 |
+
"step": 183000
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 2.4730124930930852,
|
| 2575 |
+
"grad_norm": 2.6990277767181396,
|
| 2576 |
+
"learning_rate": 8.783125115115251e-06,
|
| 2577 |
+
"loss": 0.7865,
|
| 2578 |
+
"step": 183500
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 2.479750946752739,
|
| 2582 |
+
"grad_norm": 2.4499423503875732,
|
| 2583 |
+
"learning_rate": 8.670817554121014e-06,
|
| 2584 |
+
"loss": 0.788,
|
| 2585 |
+
"step": 184000
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 2.486489400412393,
|
| 2589 |
+
"grad_norm": 2.8085732460021973,
|
| 2590 |
+
"learning_rate": 8.558509993126777e-06,
|
| 2591 |
+
"loss": 0.7843,
|
| 2592 |
+
"step": 184500
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 2.4932278540720474,
|
| 2596 |
+
"grad_norm": 2.7460155487060547,
|
| 2597 |
+
"learning_rate": 8.446202432132541e-06,
|
| 2598 |
+
"loss": 0.7837,
|
| 2599 |
+
"step": 185000
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 2.4999663077317016,
|
| 2603 |
+
"grad_norm": 2.6985998153686523,
|
| 2604 |
+
"learning_rate": 8.333894871138306e-06,
|
| 2605 |
+
"loss": 0.7796,
|
| 2606 |
+
"step": 185500
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 2.506704761391356,
|
| 2610 |
+
"grad_norm": 2.61006236076355,
|
| 2611 |
+
"learning_rate": 8.221587310144069e-06,
|
| 2612 |
+
"loss": 0.7851,
|
| 2613 |
+
"step": 186000
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 2.51344321505101,
|
| 2617 |
+
"grad_norm": 2.6244630813598633,
|
| 2618 |
+
"learning_rate": 8.109279749149833e-06,
|
| 2619 |
+
"loss": 0.7815,
|
| 2620 |
+
"step": 186500
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 2.520181668710664,
|
| 2624 |
+
"grad_norm": 2.570472240447998,
|
| 2625 |
+
"learning_rate": 7.996972188155596e-06,
|
| 2626 |
+
"loss": 0.7798,
|
| 2627 |
+
"step": 187000
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 2.5269201223703184,
|
| 2631 |
+
"grad_norm": 2.843006134033203,
|
| 2632 |
+
"learning_rate": 7.884664627161359e-06,
|
| 2633 |
+
"loss": 0.7813,
|
| 2634 |
+
"step": 187500
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 2.5336585760299726,
|
| 2638 |
+
"grad_norm": 2.826552391052246,
|
| 2639 |
+
"learning_rate": 7.772357066167123e-06,
|
| 2640 |
+
"loss": 0.7847,
|
| 2641 |
+
"step": 188000
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 2.5403970296896268,
|
| 2645 |
+
"grad_norm": 2.7578179836273193,
|
| 2646 |
+
"learning_rate": 7.660049505172886e-06,
|
| 2647 |
+
"loss": 0.7866,
|
| 2648 |
+
"step": 188500
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 2.547135483349281,
|
| 2652 |
+
"grad_norm": 2.5628886222839355,
|
| 2653 |
+
"learning_rate": 7.5477419441786505e-06,
|
| 2654 |
+
"loss": 0.7852,
|
| 2655 |
+
"step": 189000
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 2.553873937008935,
|
| 2659 |
+
"grad_norm": 2.6524205207824707,
|
| 2660 |
+
"learning_rate": 7.435434383184414e-06,
|
| 2661 |
+
"loss": 0.7814,
|
| 2662 |
+
"step": 189500
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 2.5606123906685894,
|
| 2666 |
+
"grad_norm": 2.477996349334717,
|
| 2667 |
+
"learning_rate": 7.323126822190178e-06,
|
| 2668 |
+
"loss": 0.7811,
|
| 2669 |
+
"step": 190000
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 2.5673508443282436,
|
| 2673 |
+
"grad_norm": 2.7149415016174316,
|
| 2674 |
+
"learning_rate": 7.210819261195941e-06,
|
| 2675 |
+
"loss": 0.7792,
|
| 2676 |
+
"step": 190500
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 2.5740892979878978,
|
| 2680 |
+
"grad_norm": 2.6794521808624268,
|
| 2681 |
+
"learning_rate": 7.098511700201704e-06,
|
| 2682 |
+
"loss": 0.7812,
|
| 2683 |
+
"step": 191000
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 2.580827751647552,
|
| 2687 |
+
"grad_norm": 2.5924432277679443,
|
| 2688 |
+
"learning_rate": 6.986204139207468e-06,
|
| 2689 |
+
"loss": 0.7833,
|
| 2690 |
+
"step": 191500
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 2.587566205307206,
|
| 2694 |
+
"grad_norm": 2.6516847610473633,
|
| 2695 |
+
"learning_rate": 6.873896578213232e-06,
|
| 2696 |
+
"loss": 0.7805,
|
| 2697 |
+
"step": 192000
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 2.5943046589668604,
|
| 2701 |
+
"grad_norm": 2.7591474056243896,
|
| 2702 |
+
"learning_rate": 6.761589017218996e-06,
|
| 2703 |
+
"loss": 0.7806,
|
| 2704 |
+
"step": 192500
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 2.6010431126265146,
|
| 2708 |
+
"grad_norm": 2.6870453357696533,
|
| 2709 |
+
"learning_rate": 6.64928145622476e-06,
|
| 2710 |
+
"loss": 0.7831,
|
| 2711 |
+
"step": 193000
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 2.6077815662861688,
|
| 2715 |
+
"grad_norm": 2.554856061935425,
|
| 2716 |
+
"learning_rate": 6.5369738952305235e-06,
|
| 2717 |
+
"loss": 0.7788,
|
| 2718 |
+
"step": 193500
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 2.614520019945823,
|
| 2722 |
+
"grad_norm": 3.0473780632019043,
|
| 2723 |
+
"learning_rate": 6.4246663342362855e-06,
|
| 2724 |
+
"loss": 0.7785,
|
| 2725 |
+
"step": 194000
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 2.621258473605477,
|
| 2729 |
+
"grad_norm": 2.823575973510742,
|
| 2730 |
+
"learning_rate": 6.31235877324205e-06,
|
| 2731 |
+
"loss": 0.7769,
|
| 2732 |
+
"step": 194500
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 2.6279969272651313,
|
| 2736 |
+
"grad_norm": 3.1149537563323975,
|
| 2737 |
+
"learning_rate": 6.200051212247814e-06,
|
| 2738 |
+
"loss": 0.7756,
|
| 2739 |
+
"step": 195000
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 2.6347353809247855,
|
| 2743 |
+
"grad_norm": 2.7670702934265137,
|
| 2744 |
+
"learning_rate": 6.087743651253577e-06,
|
| 2745 |
+
"loss": 0.7757,
|
| 2746 |
+
"step": 195500
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 2.6414738345844393,
|
| 2750 |
+
"grad_norm": 2.8859286308288574,
|
| 2751 |
+
"learning_rate": 5.975436090259341e-06,
|
| 2752 |
+
"loss": 0.7799,
|
| 2753 |
+
"step": 196000
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 2.6482122882440935,
|
| 2757 |
+
"grad_norm": 2.612013578414917,
|
| 2758 |
+
"learning_rate": 5.863128529265105e-06,
|
| 2759 |
+
"loss": 0.7783,
|
| 2760 |
+
"step": 196500
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 2.6549507419037477,
|
| 2764 |
+
"grad_norm": 2.4842634201049805,
|
| 2765 |
+
"learning_rate": 5.750820968270868e-06,
|
| 2766 |
+
"loss": 0.7797,
|
| 2767 |
+
"step": 197000
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 2.661689195563402,
|
| 2771 |
+
"grad_norm": 2.8483400344848633,
|
| 2772 |
+
"learning_rate": 5.638513407276632e-06,
|
| 2773 |
+
"loss": 0.7798,
|
| 2774 |
+
"step": 197500
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 2.668427649223056,
|
| 2778 |
+
"grad_norm": 2.6876285076141357,
|
| 2779 |
+
"learning_rate": 5.526205846282395e-06,
|
| 2780 |
+
"loss": 0.7763,
|
| 2781 |
+
"step": 198000
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 2.6751661028827103,
|
| 2785 |
+
"grad_norm": 2.7072014808654785,
|
| 2786 |
+
"learning_rate": 5.413898285288159e-06,
|
| 2787 |
+
"loss": 0.7777,
|
| 2788 |
+
"step": 198500
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 2.6819045565423645,
|
| 2792 |
+
"grad_norm": 2.741366386413574,
|
| 2793 |
+
"learning_rate": 5.301590724293923e-06,
|
| 2794 |
+
"loss": 0.771,
|
| 2795 |
+
"step": 199000
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 2.6886430102020187,
|
| 2799 |
+
"grad_norm": 2.619382858276367,
|
| 2800 |
+
"learning_rate": 5.189283163299686e-06,
|
| 2801 |
+
"loss": 0.7759,
|
| 2802 |
+
"step": 199500
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 2.695381463861673,
|
| 2806 |
+
"grad_norm": 2.4894635677337646,
|
| 2807 |
+
"learning_rate": 5.0769756023054495e-06,
|
| 2808 |
+
"loss": 0.7775,
|
| 2809 |
+
"step": 200000
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 2.702119917521327,
|
| 2813 |
+
"grad_norm": 2.9494383335113525,
|
| 2814 |
+
"learning_rate": 4.964668041311214e-06,
|
| 2815 |
+
"loss": 0.7725,
|
| 2816 |
+
"step": 200500
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 2.7088583711809813,
|
| 2820 |
+
"grad_norm": 2.9104485511779785,
|
| 2821 |
+
"learning_rate": 4.852360480316978e-06,
|
| 2822 |
+
"loss": 0.7756,
|
| 2823 |
+
"step": 201000
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 2.7155968248406355,
|
| 2827 |
+
"grad_norm": 3.2007381916046143,
|
| 2828 |
+
"learning_rate": 4.7400529193227405e-06,
|
| 2829 |
+
"loss": 0.7766,
|
| 2830 |
+
"step": 201500
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 2.7223352785002897,
|
| 2834 |
+
"grad_norm": 2.7596235275268555,
|
| 2835 |
+
"learning_rate": 4.627745358328504e-06,
|
| 2836 |
+
"loss": 0.7772,
|
| 2837 |
+
"step": 202000
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 2.729073732159944,
|
| 2841 |
+
"grad_norm": 2.6834709644317627,
|
| 2842 |
+
"learning_rate": 4.515437797334268e-06,
|
| 2843 |
+
"loss": 0.7749,
|
| 2844 |
+
"step": 202500
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 2.735812185819598,
|
| 2848 |
+
"grad_norm": 2.5933780670166016,
|
| 2849 |
+
"learning_rate": 4.4031302363400314e-06,
|
| 2850 |
+
"loss": 0.7719,
|
| 2851 |
+
"step": 203000
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 2.7425506394792523,
|
| 2855 |
+
"grad_norm": 2.5340027809143066,
|
| 2856 |
+
"learning_rate": 4.290822675345795e-06,
|
| 2857 |
+
"loss": 0.7741,
|
| 2858 |
+
"step": 203500
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 2.7492890931389065,
|
| 2862 |
+
"grad_norm": 2.8879024982452393,
|
| 2863 |
+
"learning_rate": 4.178515114351559e-06,
|
| 2864 |
+
"loss": 0.7716,
|
| 2865 |
+
"step": 204000
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 2.7560275467985607,
|
| 2869 |
+
"grad_norm": 3.1167986392974854,
|
| 2870 |
+
"learning_rate": 4.0662075533573224e-06,
|
| 2871 |
+
"loss": 0.7694,
|
| 2872 |
+
"step": 204500
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 2.762766000458215,
|
| 2876 |
+
"grad_norm": 2.4951884746551514,
|
| 2877 |
+
"learning_rate": 3.953899992363086e-06,
|
| 2878 |
+
"loss": 0.7765,
|
| 2879 |
+
"step": 205000
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 2.769504454117869,
|
| 2883 |
+
"grad_norm": 2.7027931213378906,
|
| 2884 |
+
"learning_rate": 3.84159243136885e-06,
|
| 2885 |
+
"loss": 0.7714,
|
| 2886 |
+
"step": 205500
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 2.7762429077775232,
|
| 2890 |
+
"grad_norm": 3.3156979084014893,
|
| 2891 |
+
"learning_rate": 3.7292848703746134e-06,
|
| 2892 |
+
"loss": 0.7781,
|
| 2893 |
+
"step": 206000
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 2.7829813614371774,
|
| 2897 |
+
"grad_norm": 2.786951780319214,
|
| 2898 |
+
"learning_rate": 3.6169773093803767e-06,
|
| 2899 |
+
"loss": 0.7728,
|
| 2900 |
+
"step": 206500
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 2.7897198150968316,
|
| 2904 |
+
"grad_norm": 2.969574451446533,
|
| 2905 |
+
"learning_rate": 3.5046697483861408e-06,
|
| 2906 |
+
"loss": 0.7738,
|
| 2907 |
+
"step": 207000
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 2.796458268756486,
|
| 2911 |
+
"grad_norm": 2.9349396228790283,
|
| 2912 |
+
"learning_rate": 3.3923621873919044e-06,
|
| 2913 |
+
"loss": 0.7706,
|
| 2914 |
+
"step": 207500
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 2.80319672241614,
|
| 2918 |
+
"grad_norm": 2.9105379581451416,
|
| 2919 |
+
"learning_rate": 3.2800546263976677e-06,
|
| 2920 |
+
"loss": 0.7726,
|
| 2921 |
+
"step": 208000
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 2.8099351760757942,
|
| 2925 |
+
"grad_norm": 2.704517364501953,
|
| 2926 |
+
"learning_rate": 3.1677470654034313e-06,
|
| 2927 |
+
"loss": 0.772,
|
| 2928 |
+
"step": 208500
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 2.8166736297354484,
|
| 2932 |
+
"grad_norm": 2.8440916538238525,
|
| 2933 |
+
"learning_rate": 3.055439504409195e-06,
|
| 2934 |
+
"loss": 0.7737,
|
| 2935 |
+
"step": 209000
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 2.8234120833951026,
|
| 2939 |
+
"grad_norm": 2.7883334159851074,
|
| 2940 |
+
"learning_rate": 2.9431319434149587e-06,
|
| 2941 |
+
"loss": 0.7761,
|
| 2942 |
+
"step": 209500
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 2.830150537054757,
|
| 2946 |
+
"grad_norm": 2.7041919231414795,
|
| 2947 |
+
"learning_rate": 2.8308243824207223e-06,
|
| 2948 |
+
"loss": 0.7685,
|
| 2949 |
+
"step": 210000
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 2.836888990714411,
|
| 2953 |
+
"grad_norm": 3.204582691192627,
|
| 2954 |
+
"learning_rate": 2.718516821426486e-06,
|
| 2955 |
+
"loss": 0.772,
|
| 2956 |
+
"step": 210500
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 2.8436274443740652,
|
| 2960 |
+
"grad_norm": 2.6497182846069336,
|
| 2961 |
+
"learning_rate": 2.6062092604322497e-06,
|
| 2962 |
+
"loss": 0.7697,
|
| 2963 |
+
"step": 211000
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 2.8503658980337194,
|
| 2967 |
+
"grad_norm": 2.7956790924072266,
|
| 2968 |
+
"learning_rate": 2.493901699438013e-06,
|
| 2969 |
+
"loss": 0.7701,
|
| 2970 |
+
"step": 211500
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 2.8571043516933736,
|
| 2974 |
+
"grad_norm": 2.669754981994629,
|
| 2975 |
+
"learning_rate": 2.3815941384437766e-06,
|
| 2976 |
+
"loss": 0.7722,
|
| 2977 |
+
"step": 212000
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 2.863842805353028,
|
| 2981 |
+
"grad_norm": 2.8052785396575928,
|
| 2982 |
+
"learning_rate": 2.2692865774495402e-06,
|
| 2983 |
+
"loss": 0.7738,
|
| 2984 |
+
"step": 212500
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 2.870581259012682,
|
| 2988 |
+
"grad_norm": 2.84078311920166,
|
| 2989 |
+
"learning_rate": 2.156979016455304e-06,
|
| 2990 |
+
"loss": 0.7634,
|
| 2991 |
+
"step": 213000
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 2.877319712672336,
|
| 2995 |
+
"grad_norm": 2.8475825786590576,
|
| 2996 |
+
"learning_rate": 2.0446714554610676e-06,
|
| 2997 |
+
"loss": 0.7701,
|
| 2998 |
+
"step": 213500
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 2.88405816633199,
|
| 3002 |
+
"grad_norm": 2.8714263439178467,
|
| 3003 |
+
"learning_rate": 1.9323638944668312e-06,
|
| 3004 |
+
"loss": 0.7675,
|
| 3005 |
+
"step": 214000
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 2.890796619991644,
|
| 3009 |
+
"grad_norm": 2.9529831409454346,
|
| 3010 |
+
"learning_rate": 1.8200563334725949e-06,
|
| 3011 |
+
"loss": 0.7634,
|
| 3012 |
+
"step": 214500
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 2.8975350736512984,
|
| 3016 |
+
"grad_norm": 2.602252960205078,
|
| 3017 |
+
"learning_rate": 1.7077487724783583e-06,
|
| 3018 |
+
"loss": 0.7743,
|
| 3019 |
+
"step": 215000
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 2.9042735273109526,
|
| 3023 |
+
"grad_norm": 2.7984066009521484,
|
| 3024 |
+
"learning_rate": 1.5954412114841222e-06,
|
| 3025 |
+
"loss": 0.7682,
|
| 3026 |
+
"step": 215500
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 2.9110119809706068,
|
| 3030 |
+
"grad_norm": 2.473423957824707,
|
| 3031 |
+
"learning_rate": 1.4831336504898857e-06,
|
| 3032 |
+
"loss": 0.7718,
|
| 3033 |
+
"step": 216000
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 2.917750434630261,
|
| 3037 |
+
"grad_norm": 2.9639880657196045,
|
| 3038 |
+
"learning_rate": 1.3708260894956493e-06,
|
| 3039 |
+
"loss": 0.7674,
|
| 3040 |
+
"step": 216500
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 2.924488888289915,
|
| 3044 |
+
"grad_norm": 2.7730565071105957,
|
| 3045 |
+
"learning_rate": 1.258518528501413e-06,
|
| 3046 |
+
"loss": 0.7689,
|
| 3047 |
+
"step": 217000
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 2.9312273419495694,
|
| 3051 |
+
"grad_norm": 2.5630087852478027,
|
| 3052 |
+
"learning_rate": 1.1462109675071764e-06,
|
| 3053 |
+
"loss": 0.7698,
|
| 3054 |
+
"step": 217500
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 2.9379657956092236,
|
| 3058 |
+
"grad_norm": 2.7039124965667725,
|
| 3059 |
+
"learning_rate": 1.03390340651294e-06,
|
| 3060 |
+
"loss": 0.7658,
|
| 3061 |
+
"step": 218000
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 2.9447042492688777,
|
| 3065 |
+
"grad_norm": 2.7059097290039062,
|
| 3066 |
+
"learning_rate": 9.215958455187038e-07,
|
| 3067 |
+
"loss": 0.7714,
|
| 3068 |
+
"step": 218500
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 2.951442702928532,
|
| 3072 |
+
"grad_norm": 2.545656681060791,
|
| 3073 |
+
"learning_rate": 8.092882845244674e-07,
|
| 3074 |
+
"loss": 0.7651,
|
| 3075 |
+
"step": 219000
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 2.958181156588186,
|
| 3079 |
+
"grad_norm": 2.602311134338379,
|
| 3080 |
+
"learning_rate": 6.96980723530231e-07,
|
| 3081 |
+
"loss": 0.7685,
|
| 3082 |
+
"step": 219500
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 2.9649196102478403,
|
| 3086 |
+
"grad_norm": 2.7936577796936035,
|
| 3087 |
+
"learning_rate": 5.846731625359947e-07,
|
| 3088 |
+
"loss": 0.7659,
|
| 3089 |
+
"step": 220000
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 2.9716580639074945,
|
| 3093 |
+
"grad_norm": 2.578080654144287,
|
| 3094 |
+
"learning_rate": 4.723656015417582e-07,
|
| 3095 |
+
"loss": 0.7708,
|
| 3096 |
+
"step": 220500
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 2.9783965175671487,
|
| 3100 |
+
"grad_norm": 2.9034383296966553,
|
| 3101 |
+
"learning_rate": 3.600580405475218e-07,
|
| 3102 |
+
"loss": 0.7693,
|
| 3103 |
+
"step": 221000
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 2.985134971226803,
|
| 3107 |
+
"grad_norm": 2.942518472671509,
|
| 3108 |
+
"learning_rate": 2.477504795532855e-07,
|
| 3109 |
+
"loss": 0.7656,
|
| 3110 |
+
"step": 221500
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 2.991873424886457,
|
| 3114 |
+
"grad_norm": 2.4221742153167725,
|
| 3115 |
+
"learning_rate": 1.3544291855904907e-07,
|
| 3116 |
+
"loss": 0.7647,
|
| 3117 |
+
"step": 222000
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 2.9986118785461113,
|
| 3121 |
+
"grad_norm": 2.536613941192627,
|
| 3122 |
+
"learning_rate": 2.3135357564812696e-08,
|
| 3123 |
+
"loss": 0.7661,
|
| 3124 |
+
"step": 222500
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 3.0,
|
| 3128 |
+
"step": 222603,
|
| 3129 |
+
"total_flos": 7.501219159976442e+18,
|
| 3130 |
+
"train_loss": 0.20371022846941,
|
| 3131 |
+
"train_runtime": 62599.538,
|
| 3132 |
+
"train_samples_per_second": 455.163,
|
| 3133 |
+
"train_steps_per_second": 3.556
|
| 3134 |
+
}
|
| 3135 |
+
],
|
| 3136 |
+
"logging_steps": 500,
|
| 3137 |
+
"max_steps": 222603,
|
| 3138 |
+
"num_input_tokens_seen": 0,
|
| 3139 |
+
"num_train_epochs": 3,
|
| 3140 |
+
"save_steps": 500,
|
| 3141 |
+
"stateful_callbacks": {
|
| 3142 |
+
"TrainerControl": {
|
| 3143 |
+
"args": {
|
| 3144 |
+
"should_epoch_stop": false,
|
| 3145 |
+
"should_evaluate": false,
|
| 3146 |
+
"should_log": false,
|
| 3147 |
+
"should_save": true,
|
| 3148 |
+
"should_training_stop": true
|
| 3149 |
+
},
|
| 3150 |
+
"attributes": {}
|
| 3151 |
+
}
|
| 3152 |
+
},
|
| 3153 |
+
"total_flos": 7.501219159976442e+18,
|
| 3154 |
+
"train_batch_size": 128,
|
| 3155 |
+
"trial_name": null,
|
| 3156 |
+
"trial_params": null
|
| 3157 |
+
}
|