End of training
Browse files- README.md +20 -2
- all_results.json +13 -13
- eval_results.json +7 -7
- train_results.json +6 -6
- trainer_state.json +4311 -15
README.md
CHANGED
|
@@ -2,9 +2,24 @@
|
|
| 2 |
base_model: zaenalium/indonesia-distilgpt2
|
| 3 |
tags:
|
| 4 |
- generated_from_trainer
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
model-index:
|
| 6 |
- name: indonesia-distilgpt2
|
| 7 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
---
|
| 9 |
|
| 10 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -12,7 +27,10 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 12 |
|
| 13 |
# indonesia-distilgpt2
|
| 14 |
|
| 15 |
-
This model is a fine-tuned version of [zaenalium/indonesia-distilgpt2](https://huggingface.co/zaenalium/indonesia-distilgpt2) on
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
## Model description
|
| 18 |
|
|
|
|
| 2 |
base_model: zaenalium/indonesia-distilgpt2
|
| 3 |
tags:
|
| 4 |
- generated_from_trainer
|
| 5 |
+
datasets:
|
| 6 |
+
- wikimedia/wikipedia
|
| 7 |
+
metrics:
|
| 8 |
+
- accuracy
|
| 9 |
model-index:
|
| 10 |
- name: indonesia-distilgpt2
|
| 11 |
+
results:
|
| 12 |
+
- task:
|
| 13 |
+
name: Causal Language Modeling
|
| 14 |
+
type: text-generation
|
| 15 |
+
dataset:
|
| 16 |
+
name: wikimedia/wikipedia 20231101.id
|
| 17 |
+
type: wikimedia/wikipedia
|
| 18 |
+
args: 20231101.id
|
| 19 |
+
metrics:
|
| 20 |
+
- name: Accuracy
|
| 21 |
+
type: accuracy
|
| 22 |
+
value: 0.47740403110973484
|
| 23 |
---
|
| 24 |
|
| 25 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 27 |
|
| 28 |
# indonesia-distilgpt2
|
| 29 |
|
| 30 |
+
This model is a fine-tuned version of [zaenalium/indonesia-distilgpt2](https://huggingface.co/zaenalium/indonesia-distilgpt2) on the wikimedia/wikipedia 20231101.id dataset.
|
| 31 |
+
It achieves the following results on the evaluation set:
|
| 32 |
+
- Loss: 2.5680
|
| 33 |
+
- Accuracy: 0.4774
|
| 34 |
|
| 35 |
## Model description
|
| 36 |
|
all_results.json
CHANGED
|
@@ -1,16 +1,16 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 2.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss":
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples":
|
| 7 |
-
"eval_samples_per_second":
|
| 8 |
-
"eval_steps_per_second":
|
| 9 |
-
"perplexity":
|
| 10 |
-
"total_flos":
|
| 11 |
-
"train_loss":
|
| 12 |
-
"train_runtime":
|
| 13 |
-
"train_samples":
|
| 14 |
-
"train_samples_per_second":
|
| 15 |
-
"train_steps_per_second":
|
| 16 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 2.0,
|
| 3 |
+
"eval_accuracy": 0.47740403110973484,
|
| 4 |
+
"eval_loss": 2.567959785461426,
|
| 5 |
+
"eval_runtime": 4355.6296,
|
| 6 |
+
"eval_samples": 41158,
|
| 7 |
+
"eval_samples_per_second": 9.449,
|
| 8 |
+
"eval_steps_per_second": 2.362,
|
| 9 |
+
"perplexity": 13.03919453055406,
|
| 10 |
+
"total_flos": 1.502524253059154e+17,
|
| 11 |
+
"train_loss": 2.352252098808129,
|
| 12 |
+
"train_runtime": 164976.46,
|
| 13 |
+
"train_samples": 287513,
|
| 14 |
+
"train_samples_per_second": 3.486,
|
| 15 |
+
"train_steps_per_second": 0.871
|
| 16 |
}
|
eval_results.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 2.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss":
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples":
|
| 7 |
-
"eval_samples_per_second":
|
| 8 |
-
"eval_steps_per_second":
|
| 9 |
-
"perplexity":
|
| 10 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 2.0,
|
| 3 |
+
"eval_accuracy": 0.47740403110973484,
|
| 4 |
+
"eval_loss": 2.567959785461426,
|
| 5 |
+
"eval_runtime": 4355.6296,
|
| 6 |
+
"eval_samples": 41158,
|
| 7 |
+
"eval_samples_per_second": 9.449,
|
| 8 |
+
"eval_steps_per_second": 2.362,
|
| 9 |
+
"perplexity": 13.03919453055406
|
| 10 |
}
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 2.0,
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss":
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples":
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second":
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 2.0,
|
| 3 |
+
"total_flos": 1.502524253059154e+17,
|
| 4 |
+
"train_loss": 2.352252098808129,
|
| 5 |
+
"train_runtime": 164976.46,
|
| 6 |
+
"train_samples": 287513,
|
| 7 |
+
"train_samples_per_second": 3.486,
|
| 8 |
+
"train_steps_per_second": 0.871
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -3,39 +3,4335 @@
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 2.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 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.
|
| 13 |
-
"learning_rate":
|
| 14 |
-
"loss": 3.
|
| 15 |
"step": 200
|
| 16 |
},
|
| 17 |
{
|
| 18 |
-
"epoch":
|
| 19 |
-
"learning_rate":
|
| 20 |
-
"loss":
|
| 21 |
"step": 400
|
| 22 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
{
|
| 24 |
"epoch": 2.0,
|
| 25 |
-
"step":
|
| 26 |
-
"total_flos":
|
| 27 |
-
"train_loss":
|
| 28 |
-
"train_runtime":
|
| 29 |
-
"train_samples_per_second":
|
| 30 |
-
"train_steps_per_second":
|
| 31 |
}
|
| 32 |
],
|
| 33 |
"logging_steps": 200,
|
| 34 |
-
"max_steps":
|
| 35 |
"num_input_tokens_seen": 0,
|
| 36 |
"num_train_epochs": 2,
|
| 37 |
"save_steps": 500,
|
| 38 |
-
"total_flos":
|
| 39 |
"train_batch_size": 4,
|
| 40 |
"trial_name": null,
|
| 41 |
"trial_params": null
|
|
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 2.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 143758,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 4.993043865384883e-05,
|
| 14 |
+
"loss": 3.1775,
|
| 15 |
"step": 200
|
| 16 |
},
|
| 17 |
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 4.986087730769766e-05,
|
| 20 |
+
"loss": 2.9882,
|
| 21 |
"step": 400
|
| 22 |
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 4.979131596154649e-05,
|
| 26 |
+
"loss": 2.932,
|
| 27 |
+
"step": 600
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.01,
|
| 31 |
+
"learning_rate": 4.9721754615395325e-05,
|
| 32 |
+
"loss": 2.8774,
|
| 33 |
+
"step": 800
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.01,
|
| 37 |
+
"learning_rate": 4.965219326924415e-05,
|
| 38 |
+
"loss": 2.9227,
|
| 39 |
+
"step": 1000
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02,
|
| 43 |
+
"learning_rate": 4.9582631923092976e-05,
|
| 44 |
+
"loss": 2.9011,
|
| 45 |
+
"step": 1200
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.02,
|
| 49 |
+
"learning_rate": 4.9513070576941804e-05,
|
| 50 |
+
"loss": 2.8233,
|
| 51 |
+
"step": 1400
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.02,
|
| 55 |
+
"learning_rate": 4.944350923079063e-05,
|
| 56 |
+
"loss": 2.859,
|
| 57 |
+
"step": 1600
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.03,
|
| 61 |
+
"learning_rate": 4.937394788463946e-05,
|
| 62 |
+
"loss": 2.8269,
|
| 63 |
+
"step": 1800
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.03,
|
| 67 |
+
"learning_rate": 4.93043865384883e-05,
|
| 68 |
+
"loss": 2.82,
|
| 69 |
+
"step": 2000
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.03,
|
| 73 |
+
"learning_rate": 4.9234825192337127e-05,
|
| 74 |
+
"loss": 2.7556,
|
| 75 |
+
"step": 2200
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.03,
|
| 79 |
+
"learning_rate": 4.9165263846185955e-05,
|
| 80 |
+
"loss": 2.7388,
|
| 81 |
+
"step": 2400
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.04,
|
| 85 |
+
"learning_rate": 4.9095702500034784e-05,
|
| 86 |
+
"loss": 2.8441,
|
| 87 |
+
"step": 2600
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.04,
|
| 91 |
+
"learning_rate": 4.902614115388361e-05,
|
| 92 |
+
"loss": 2.7828,
|
| 93 |
+
"step": 2800
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.04,
|
| 97 |
+
"learning_rate": 4.895657980773244e-05,
|
| 98 |
+
"loss": 2.7528,
|
| 99 |
+
"step": 3000
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.04,
|
| 103 |
+
"learning_rate": 4.888701846158127e-05,
|
| 104 |
+
"loss": 2.7744,
|
| 105 |
+
"step": 3200
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.05,
|
| 109 |
+
"learning_rate": 4.88174571154301e-05,
|
| 110 |
+
"loss": 2.7752,
|
| 111 |
+
"step": 3400
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.05,
|
| 115 |
+
"learning_rate": 4.874789576927893e-05,
|
| 116 |
+
"loss": 2.7288,
|
| 117 |
+
"step": 3600
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.05,
|
| 121 |
+
"learning_rate": 4.867833442312776e-05,
|
| 122 |
+
"loss": 2.7043,
|
| 123 |
+
"step": 3800
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.06,
|
| 127 |
+
"learning_rate": 4.8608773076976586e-05,
|
| 128 |
+
"loss": 2.7296,
|
| 129 |
+
"step": 4000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.06,
|
| 133 |
+
"learning_rate": 4.8539211730825415e-05,
|
| 134 |
+
"loss": 2.6831,
|
| 135 |
+
"step": 4200
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.06,
|
| 139 |
+
"learning_rate": 4.846965038467425e-05,
|
| 140 |
+
"loss": 2.7519,
|
| 141 |
+
"step": 4400
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.06,
|
| 145 |
+
"learning_rate": 4.840008903852308e-05,
|
| 146 |
+
"loss": 2.6923,
|
| 147 |
+
"step": 4600
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.07,
|
| 151 |
+
"learning_rate": 4.833052769237191e-05,
|
| 152 |
+
"loss": 2.6681,
|
| 153 |
+
"step": 4800
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.07,
|
| 157 |
+
"learning_rate": 4.826096634622073e-05,
|
| 158 |
+
"loss": 2.6918,
|
| 159 |
+
"step": 5000
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.07,
|
| 163 |
+
"learning_rate": 4.819140500006956e-05,
|
| 164 |
+
"loss": 2.6827,
|
| 165 |
+
"step": 5200
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.08,
|
| 169 |
+
"learning_rate": 4.812184365391839e-05,
|
| 170 |
+
"loss": 2.6477,
|
| 171 |
+
"step": 5400
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.08,
|
| 175 |
+
"learning_rate": 4.8052282307767224e-05,
|
| 176 |
+
"loss": 2.6638,
|
| 177 |
+
"step": 5600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.08,
|
| 181 |
+
"learning_rate": 4.798272096161605e-05,
|
| 182 |
+
"loss": 2.7062,
|
| 183 |
+
"step": 5800
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.08,
|
| 187 |
+
"learning_rate": 4.791315961546488e-05,
|
| 188 |
+
"loss": 2.6954,
|
| 189 |
+
"step": 6000
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.09,
|
| 193 |
+
"learning_rate": 4.784359826931371e-05,
|
| 194 |
+
"loss": 2.7133,
|
| 195 |
+
"step": 6200
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.09,
|
| 199 |
+
"learning_rate": 4.777403692316254e-05,
|
| 200 |
+
"loss": 2.693,
|
| 201 |
+
"step": 6400
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.09,
|
| 205 |
+
"learning_rate": 4.770447557701137e-05,
|
| 206 |
+
"loss": 2.632,
|
| 207 |
+
"step": 6600
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.09,
|
| 211 |
+
"learning_rate": 4.76349142308602e-05,
|
| 212 |
+
"loss": 2.6715,
|
| 213 |
+
"step": 6800
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.1,
|
| 217 |
+
"learning_rate": 4.7565352884709026e-05,
|
| 218 |
+
"loss": 2.6947,
|
| 219 |
+
"step": 7000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.1,
|
| 223 |
+
"learning_rate": 4.7495791538557855e-05,
|
| 224 |
+
"loss": 2.6396,
|
| 225 |
+
"step": 7200
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.1,
|
| 229 |
+
"learning_rate": 4.7426230192406684e-05,
|
| 230 |
+
"loss": 2.6506,
|
| 231 |
+
"step": 7400
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.11,
|
| 235 |
+
"learning_rate": 4.735666884625551e-05,
|
| 236 |
+
"loss": 2.6204,
|
| 237 |
+
"step": 7600
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.11,
|
| 241 |
+
"learning_rate": 4.728710750010434e-05,
|
| 242 |
+
"loss": 2.6298,
|
| 243 |
+
"step": 7800
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.11,
|
| 247 |
+
"learning_rate": 4.721754615395318e-05,
|
| 248 |
+
"loss": 2.5967,
|
| 249 |
+
"step": 8000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.11,
|
| 253 |
+
"learning_rate": 4.7147984807802006e-05,
|
| 254 |
+
"loss": 2.6396,
|
| 255 |
+
"step": 8200
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.12,
|
| 259 |
+
"learning_rate": 4.7078423461650835e-05,
|
| 260 |
+
"loss": 2.601,
|
| 261 |
+
"step": 8400
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.12,
|
| 265 |
+
"learning_rate": 4.7008862115499664e-05,
|
| 266 |
+
"loss": 2.6289,
|
| 267 |
+
"step": 8600
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.12,
|
| 271 |
+
"learning_rate": 4.693930076934849e-05,
|
| 272 |
+
"loss": 2.6143,
|
| 273 |
+
"step": 8800
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.13,
|
| 277 |
+
"learning_rate": 4.6869739423197315e-05,
|
| 278 |
+
"loss": 2.6672,
|
| 279 |
+
"step": 9000
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.13,
|
| 283 |
+
"learning_rate": 4.680017807704615e-05,
|
| 284 |
+
"loss": 2.6145,
|
| 285 |
+
"step": 9200
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.13,
|
| 289 |
+
"learning_rate": 4.673061673089498e-05,
|
| 290 |
+
"loss": 2.626,
|
| 291 |
+
"step": 9400
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.13,
|
| 295 |
+
"learning_rate": 4.666105538474381e-05,
|
| 296 |
+
"loss": 2.6622,
|
| 297 |
+
"step": 9600
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.14,
|
| 301 |
+
"learning_rate": 4.659149403859264e-05,
|
| 302 |
+
"loss": 2.578,
|
| 303 |
+
"step": 9800
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.14,
|
| 307 |
+
"learning_rate": 4.6521932692441466e-05,
|
| 308 |
+
"loss": 2.6476,
|
| 309 |
+
"step": 10000
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.14,
|
| 313 |
+
"learning_rate": 4.6452371346290294e-05,
|
| 314 |
+
"loss": 2.6051,
|
| 315 |
+
"step": 10200
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.14,
|
| 319 |
+
"learning_rate": 4.638281000013912e-05,
|
| 320 |
+
"loss": 2.5568,
|
| 321 |
+
"step": 10400
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.15,
|
| 325 |
+
"learning_rate": 4.631324865398796e-05,
|
| 326 |
+
"loss": 2.5498,
|
| 327 |
+
"step": 10600
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.15,
|
| 331 |
+
"learning_rate": 4.624368730783679e-05,
|
| 332 |
+
"loss": 2.5283,
|
| 333 |
+
"step": 10800
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.15,
|
| 337 |
+
"learning_rate": 4.617412596168561e-05,
|
| 338 |
+
"loss": 2.615,
|
| 339 |
+
"step": 11000
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.16,
|
| 343 |
+
"learning_rate": 4.610456461553444e-05,
|
| 344 |
+
"loss": 2.6121,
|
| 345 |
+
"step": 11200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.16,
|
| 349 |
+
"learning_rate": 4.603500326938327e-05,
|
| 350 |
+
"loss": 2.535,
|
| 351 |
+
"step": 11400
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.16,
|
| 355 |
+
"learning_rate": 4.59654419232321e-05,
|
| 356 |
+
"loss": 2.584,
|
| 357 |
+
"step": 11600
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.16,
|
| 361 |
+
"learning_rate": 4.589588057708093e-05,
|
| 362 |
+
"loss": 2.5619,
|
| 363 |
+
"step": 11800
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.17,
|
| 367 |
+
"learning_rate": 4.582631923092976e-05,
|
| 368 |
+
"loss": 2.581,
|
| 369 |
+
"step": 12000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.17,
|
| 373 |
+
"learning_rate": 4.575675788477859e-05,
|
| 374 |
+
"loss": 2.5513,
|
| 375 |
+
"step": 12200
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.17,
|
| 379 |
+
"learning_rate": 4.568719653862742e-05,
|
| 380 |
+
"loss": 2.5844,
|
| 381 |
+
"step": 12400
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.18,
|
| 385 |
+
"learning_rate": 4.561763519247625e-05,
|
| 386 |
+
"loss": 2.6033,
|
| 387 |
+
"step": 12600
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.18,
|
| 391 |
+
"learning_rate": 4.5548073846325076e-05,
|
| 392 |
+
"loss": 2.4988,
|
| 393 |
+
"step": 12800
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.18,
|
| 397 |
+
"learning_rate": 4.5478512500173905e-05,
|
| 398 |
+
"loss": 2.5467,
|
| 399 |
+
"step": 13000
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.18,
|
| 403 |
+
"learning_rate": 4.5408951154022734e-05,
|
| 404 |
+
"loss": 2.5825,
|
| 405 |
+
"step": 13200
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.19,
|
| 409 |
+
"learning_rate": 4.533938980787156e-05,
|
| 410 |
+
"loss": 2.5732,
|
| 411 |
+
"step": 13400
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.19,
|
| 415 |
+
"learning_rate": 4.526982846172039e-05,
|
| 416 |
+
"loss": 2.572,
|
| 417 |
+
"step": 13600
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.19,
|
| 421 |
+
"learning_rate": 4.520026711556922e-05,
|
| 422 |
+
"loss": 2.4804,
|
| 423 |
+
"step": 13800
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.19,
|
| 427 |
+
"learning_rate": 4.513070576941805e-05,
|
| 428 |
+
"loss": 2.5576,
|
| 429 |
+
"step": 14000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.2,
|
| 433 |
+
"learning_rate": 4.5061144423266885e-05,
|
| 434 |
+
"loss": 2.5474,
|
| 435 |
+
"step": 14200
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.2,
|
| 439 |
+
"learning_rate": 4.4991583077115714e-05,
|
| 440 |
+
"loss": 2.5262,
|
| 441 |
+
"step": 14400
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.2,
|
| 445 |
+
"learning_rate": 4.492202173096454e-05,
|
| 446 |
+
"loss": 2.5679,
|
| 447 |
+
"step": 14600
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.21,
|
| 451 |
+
"learning_rate": 4.485246038481337e-05,
|
| 452 |
+
"loss": 2.5541,
|
| 453 |
+
"step": 14800
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.21,
|
| 457 |
+
"learning_rate": 4.4782899038662194e-05,
|
| 458 |
+
"loss": 2.4937,
|
| 459 |
+
"step": 15000
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.21,
|
| 463 |
+
"learning_rate": 4.471333769251102e-05,
|
| 464 |
+
"loss": 2.5643,
|
| 465 |
+
"step": 15200
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.21,
|
| 469 |
+
"learning_rate": 4.464377634635986e-05,
|
| 470 |
+
"loss": 2.5978,
|
| 471 |
+
"step": 15400
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.22,
|
| 475 |
+
"learning_rate": 4.457421500020869e-05,
|
| 476 |
+
"loss": 2.4727,
|
| 477 |
+
"step": 15600
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.22,
|
| 481 |
+
"learning_rate": 4.4504653654057516e-05,
|
| 482 |
+
"loss": 2.613,
|
| 483 |
+
"step": 15800
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.22,
|
| 487 |
+
"learning_rate": 4.4435092307906345e-05,
|
| 488 |
+
"loss": 2.5453,
|
| 489 |
+
"step": 16000
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.23,
|
| 493 |
+
"learning_rate": 4.4365530961755174e-05,
|
| 494 |
+
"loss": 2.5362,
|
| 495 |
+
"step": 16200
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.23,
|
| 499 |
+
"learning_rate": 4.4295969615604e-05,
|
| 500 |
+
"loss": 2.5296,
|
| 501 |
+
"step": 16400
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.23,
|
| 505 |
+
"learning_rate": 4.422640826945283e-05,
|
| 506 |
+
"loss": 2.5282,
|
| 507 |
+
"step": 16600
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.23,
|
| 511 |
+
"learning_rate": 4.415684692330167e-05,
|
| 512 |
+
"loss": 2.4973,
|
| 513 |
+
"step": 16800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.24,
|
| 517 |
+
"learning_rate": 4.408728557715049e-05,
|
| 518 |
+
"loss": 2.5563,
|
| 519 |
+
"step": 17000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.24,
|
| 523 |
+
"learning_rate": 4.401772423099932e-05,
|
| 524 |
+
"loss": 2.5514,
|
| 525 |
+
"step": 17200
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.24,
|
| 529 |
+
"learning_rate": 4.394816288484815e-05,
|
| 530 |
+
"loss": 2.5327,
|
| 531 |
+
"step": 17400
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.24,
|
| 535 |
+
"learning_rate": 4.3878601538696976e-05,
|
| 536 |
+
"loss": 2.5251,
|
| 537 |
+
"step": 17600
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.25,
|
| 541 |
+
"learning_rate": 4.380904019254581e-05,
|
| 542 |
+
"loss": 2.5695,
|
| 543 |
+
"step": 17800
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.25,
|
| 547 |
+
"learning_rate": 4.373947884639464e-05,
|
| 548 |
+
"loss": 2.4936,
|
| 549 |
+
"step": 18000
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.25,
|
| 553 |
+
"learning_rate": 4.366991750024347e-05,
|
| 554 |
+
"loss": 2.516,
|
| 555 |
+
"step": 18200
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.26,
|
| 559 |
+
"learning_rate": 4.36003561540923e-05,
|
| 560 |
+
"loss": 2.5568,
|
| 561 |
+
"step": 18400
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.26,
|
| 565 |
+
"learning_rate": 4.353079480794113e-05,
|
| 566 |
+
"loss": 2.5253,
|
| 567 |
+
"step": 18600
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.26,
|
| 571 |
+
"learning_rate": 4.346123346178995e-05,
|
| 572 |
+
"loss": 2.4991,
|
| 573 |
+
"step": 18800
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.26,
|
| 577 |
+
"learning_rate": 4.3391672115638785e-05,
|
| 578 |
+
"loss": 2.5029,
|
| 579 |
+
"step": 19000
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.27,
|
| 583 |
+
"learning_rate": 4.332211076948761e-05,
|
| 584 |
+
"loss": 2.5178,
|
| 585 |
+
"step": 19200
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.27,
|
| 589 |
+
"learning_rate": 4.325254942333644e-05,
|
| 590 |
+
"loss": 2.4964,
|
| 591 |
+
"step": 19400
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.27,
|
| 595 |
+
"learning_rate": 4.318298807718527e-05,
|
| 596 |
+
"loss": 2.4339,
|
| 597 |
+
"step": 19600
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.28,
|
| 601 |
+
"learning_rate": 4.31134267310341e-05,
|
| 602 |
+
"loss": 2.4874,
|
| 603 |
+
"step": 19800
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.28,
|
| 607 |
+
"learning_rate": 4.304386538488293e-05,
|
| 608 |
+
"loss": 2.4894,
|
| 609 |
+
"step": 20000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.28,
|
| 613 |
+
"learning_rate": 4.297430403873176e-05,
|
| 614 |
+
"loss": 2.5352,
|
| 615 |
+
"step": 20200
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.28,
|
| 619 |
+
"learning_rate": 4.290474269258059e-05,
|
| 620 |
+
"loss": 2.5251,
|
| 621 |
+
"step": 20400
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.29,
|
| 625 |
+
"learning_rate": 4.283518134642942e-05,
|
| 626 |
+
"loss": 2.4545,
|
| 627 |
+
"step": 20600
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.29,
|
| 631 |
+
"learning_rate": 4.2765620000278244e-05,
|
| 632 |
+
"loss": 2.4831,
|
| 633 |
+
"step": 20800
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.29,
|
| 637 |
+
"learning_rate": 4.269605865412707e-05,
|
| 638 |
+
"loss": 2.5145,
|
| 639 |
+
"step": 21000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.29,
|
| 643 |
+
"learning_rate": 4.26264973079759e-05,
|
| 644 |
+
"loss": 2.4543,
|
| 645 |
+
"step": 21200
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.3,
|
| 649 |
+
"learning_rate": 4.255693596182474e-05,
|
| 650 |
+
"loss": 2.5056,
|
| 651 |
+
"step": 21400
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 0.3,
|
| 655 |
+
"learning_rate": 4.2487374615673566e-05,
|
| 656 |
+
"loss": 2.4933,
|
| 657 |
+
"step": 21600
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.3,
|
| 661 |
+
"learning_rate": 4.2417813269522395e-05,
|
| 662 |
+
"loss": 2.4786,
|
| 663 |
+
"step": 21800
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 0.31,
|
| 667 |
+
"learning_rate": 4.2348251923371224e-05,
|
| 668 |
+
"loss": 2.3971,
|
| 669 |
+
"step": 22000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.31,
|
| 673 |
+
"learning_rate": 4.227869057722005e-05,
|
| 674 |
+
"loss": 2.4759,
|
| 675 |
+
"step": 22200
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.31,
|
| 679 |
+
"learning_rate": 4.220912923106888e-05,
|
| 680 |
+
"loss": 2.4637,
|
| 681 |
+
"step": 22400
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.31,
|
| 685 |
+
"learning_rate": 4.213956788491771e-05,
|
| 686 |
+
"loss": 2.5033,
|
| 687 |
+
"step": 22600
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 0.32,
|
| 691 |
+
"learning_rate": 4.207000653876654e-05,
|
| 692 |
+
"loss": 2.5021,
|
| 693 |
+
"step": 22800
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.32,
|
| 697 |
+
"learning_rate": 4.200044519261537e-05,
|
| 698 |
+
"loss": 2.4136,
|
| 699 |
+
"step": 23000
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.32,
|
| 703 |
+
"learning_rate": 4.19308838464642e-05,
|
| 704 |
+
"loss": 2.5604,
|
| 705 |
+
"step": 23200
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.33,
|
| 709 |
+
"learning_rate": 4.1861322500313026e-05,
|
| 710 |
+
"loss": 2.5054,
|
| 711 |
+
"step": 23400
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.33,
|
| 715 |
+
"learning_rate": 4.1791761154161855e-05,
|
| 716 |
+
"loss": 2.4761,
|
| 717 |
+
"step": 23600
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.33,
|
| 721 |
+
"learning_rate": 4.1722199808010684e-05,
|
| 722 |
+
"loss": 2.4802,
|
| 723 |
+
"step": 23800
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.33,
|
| 727 |
+
"learning_rate": 4.165263846185952e-05,
|
| 728 |
+
"loss": 2.5241,
|
| 729 |
+
"step": 24000
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.34,
|
| 733 |
+
"learning_rate": 4.158307711570835e-05,
|
| 734 |
+
"loss": 2.4519,
|
| 735 |
+
"step": 24200
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 0.34,
|
| 739 |
+
"learning_rate": 4.151351576955718e-05,
|
| 740 |
+
"loss": 2.487,
|
| 741 |
+
"step": 24400
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.34,
|
| 745 |
+
"learning_rate": 4.1443954423406006e-05,
|
| 746 |
+
"loss": 2.4719,
|
| 747 |
+
"step": 24600
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.35,
|
| 751 |
+
"learning_rate": 4.137439307725483e-05,
|
| 752 |
+
"loss": 2.5132,
|
| 753 |
+
"step": 24800
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.35,
|
| 757 |
+
"learning_rate": 4.130483173110366e-05,
|
| 758 |
+
"loss": 2.4432,
|
| 759 |
+
"step": 25000
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.35,
|
| 763 |
+
"learning_rate": 4.123527038495249e-05,
|
| 764 |
+
"loss": 2.448,
|
| 765 |
+
"step": 25200
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.35,
|
| 769 |
+
"learning_rate": 4.116570903880132e-05,
|
| 770 |
+
"loss": 2.4659,
|
| 771 |
+
"step": 25400
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 0.36,
|
| 775 |
+
"learning_rate": 4.109614769265015e-05,
|
| 776 |
+
"loss": 2.4596,
|
| 777 |
+
"step": 25600
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.36,
|
| 781 |
+
"learning_rate": 4.102658634649898e-05,
|
| 782 |
+
"loss": 2.473,
|
| 783 |
+
"step": 25800
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.36,
|
| 787 |
+
"learning_rate": 4.095702500034781e-05,
|
| 788 |
+
"loss": 2.4812,
|
| 789 |
+
"step": 26000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.36,
|
| 793 |
+
"learning_rate": 4.088746365419664e-05,
|
| 794 |
+
"loss": 2.43,
|
| 795 |
+
"step": 26200
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.37,
|
| 799 |
+
"learning_rate": 4.081790230804547e-05,
|
| 800 |
+
"loss": 2.4743,
|
| 801 |
+
"step": 26400
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.37,
|
| 805 |
+
"learning_rate": 4.07483409618943e-05,
|
| 806 |
+
"loss": 2.4693,
|
| 807 |
+
"step": 26600
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.37,
|
| 811 |
+
"learning_rate": 4.0678779615743124e-05,
|
| 812 |
+
"loss": 2.4932,
|
| 813 |
+
"step": 26800
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.38,
|
| 817 |
+
"learning_rate": 4.060921826959195e-05,
|
| 818 |
+
"loss": 2.4459,
|
| 819 |
+
"step": 27000
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.38,
|
| 823 |
+
"learning_rate": 4.053965692344078e-05,
|
| 824 |
+
"loss": 2.4123,
|
| 825 |
+
"step": 27200
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.38,
|
| 829 |
+
"learning_rate": 4.047009557728961e-05,
|
| 830 |
+
"loss": 2.4364,
|
| 831 |
+
"step": 27400
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.38,
|
| 835 |
+
"learning_rate": 4.0400534231138446e-05,
|
| 836 |
+
"loss": 2.4428,
|
| 837 |
+
"step": 27600
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.39,
|
| 841 |
+
"learning_rate": 4.0330972884987275e-05,
|
| 842 |
+
"loss": 2.4157,
|
| 843 |
+
"step": 27800
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.39,
|
| 847 |
+
"learning_rate": 4.0261411538836103e-05,
|
| 848 |
+
"loss": 2.4145,
|
| 849 |
+
"step": 28000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.39,
|
| 853 |
+
"learning_rate": 4.019185019268493e-05,
|
| 854 |
+
"loss": 2.5131,
|
| 855 |
+
"step": 28200
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.4,
|
| 859 |
+
"learning_rate": 4.012228884653376e-05,
|
| 860 |
+
"loss": 2.4509,
|
| 861 |
+
"step": 28400
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.4,
|
| 865 |
+
"learning_rate": 4.005272750038259e-05,
|
| 866 |
+
"loss": 2.4781,
|
| 867 |
+
"step": 28600
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.4,
|
| 871 |
+
"learning_rate": 3.998316615423142e-05,
|
| 872 |
+
"loss": 2.428,
|
| 873 |
+
"step": 28800
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.4,
|
| 877 |
+
"learning_rate": 3.991360480808025e-05,
|
| 878 |
+
"loss": 2.4402,
|
| 879 |
+
"step": 29000
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.41,
|
| 883 |
+
"learning_rate": 3.9844043461929077e-05,
|
| 884 |
+
"loss": 2.3886,
|
| 885 |
+
"step": 29200
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.41,
|
| 889 |
+
"learning_rate": 3.9774482115777905e-05,
|
| 890 |
+
"loss": 2.4329,
|
| 891 |
+
"step": 29400
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.41,
|
| 895 |
+
"learning_rate": 3.9704920769626734e-05,
|
| 896 |
+
"loss": 2.4038,
|
| 897 |
+
"step": 29600
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.41,
|
| 901 |
+
"learning_rate": 3.963535942347556e-05,
|
| 902 |
+
"loss": 2.3857,
|
| 903 |
+
"step": 29800
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.42,
|
| 907 |
+
"learning_rate": 3.956579807732439e-05,
|
| 908 |
+
"loss": 2.3806,
|
| 909 |
+
"step": 30000
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.42,
|
| 913 |
+
"learning_rate": 3.949623673117323e-05,
|
| 914 |
+
"loss": 2.4243,
|
| 915 |
+
"step": 30200
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.42,
|
| 919 |
+
"learning_rate": 3.9426675385022057e-05,
|
| 920 |
+
"loss": 2.4545,
|
| 921 |
+
"step": 30400
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.43,
|
| 925 |
+
"learning_rate": 3.9357114038870885e-05,
|
| 926 |
+
"loss": 2.5031,
|
| 927 |
+
"step": 30600
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.43,
|
| 931 |
+
"learning_rate": 3.928755269271971e-05,
|
| 932 |
+
"loss": 2.4046,
|
| 933 |
+
"step": 30800
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.43,
|
| 937 |
+
"learning_rate": 3.9217991346568536e-05,
|
| 938 |
+
"loss": 2.4326,
|
| 939 |
+
"step": 31000
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.43,
|
| 943 |
+
"learning_rate": 3.914843000041737e-05,
|
| 944 |
+
"loss": 2.4118,
|
| 945 |
+
"step": 31200
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.44,
|
| 949 |
+
"learning_rate": 3.90788686542662e-05,
|
| 950 |
+
"loss": 2.4556,
|
| 951 |
+
"step": 31400
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 0.44,
|
| 955 |
+
"learning_rate": 3.900930730811503e-05,
|
| 956 |
+
"loss": 2.4579,
|
| 957 |
+
"step": 31600
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 0.44,
|
| 961 |
+
"learning_rate": 3.893974596196386e-05,
|
| 962 |
+
"loss": 2.3932,
|
| 963 |
+
"step": 31800
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 0.45,
|
| 967 |
+
"learning_rate": 3.887018461581269e-05,
|
| 968 |
+
"loss": 2.4281,
|
| 969 |
+
"step": 32000
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.45,
|
| 973 |
+
"learning_rate": 3.8800623269661516e-05,
|
| 974 |
+
"loss": 2.4362,
|
| 975 |
+
"step": 32200
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.45,
|
| 979 |
+
"learning_rate": 3.8731061923510345e-05,
|
| 980 |
+
"loss": 2.4217,
|
| 981 |
+
"step": 32400
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.45,
|
| 985 |
+
"learning_rate": 3.866150057735918e-05,
|
| 986 |
+
"loss": 2.4305,
|
| 987 |
+
"step": 32600
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.46,
|
| 991 |
+
"learning_rate": 3.8591939231208e-05,
|
| 992 |
+
"loss": 2.441,
|
| 993 |
+
"step": 32800
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 0.46,
|
| 997 |
+
"learning_rate": 3.852237788505683e-05,
|
| 998 |
+
"loss": 2.422,
|
| 999 |
+
"step": 33000
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.46,
|
| 1003 |
+
"learning_rate": 3.845281653890566e-05,
|
| 1004 |
+
"loss": 2.3997,
|
| 1005 |
+
"step": 33200
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 0.46,
|
| 1009 |
+
"learning_rate": 3.838325519275449e-05,
|
| 1010 |
+
"loss": 2.3749,
|
| 1011 |
+
"step": 33400
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.47,
|
| 1015 |
+
"learning_rate": 3.831369384660332e-05,
|
| 1016 |
+
"loss": 2.4073,
|
| 1017 |
+
"step": 33600
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.47,
|
| 1021 |
+
"learning_rate": 3.8244132500452154e-05,
|
| 1022 |
+
"loss": 2.4585,
|
| 1023 |
+
"step": 33800
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 0.47,
|
| 1027 |
+
"learning_rate": 3.817457115430098e-05,
|
| 1028 |
+
"loss": 2.4215,
|
| 1029 |
+
"step": 34000
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.48,
|
| 1033 |
+
"learning_rate": 3.810500980814981e-05,
|
| 1034 |
+
"loss": 2.414,
|
| 1035 |
+
"step": 34200
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.48,
|
| 1039 |
+
"learning_rate": 3.803544846199864e-05,
|
| 1040 |
+
"loss": 2.4223,
|
| 1041 |
+
"step": 34400
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 0.48,
|
| 1045 |
+
"learning_rate": 3.796588711584746e-05,
|
| 1046 |
+
"loss": 2.4012,
|
| 1047 |
+
"step": 34600
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 0.48,
|
| 1051 |
+
"learning_rate": 3.78963257696963e-05,
|
| 1052 |
+
"loss": 2.422,
|
| 1053 |
+
"step": 34800
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.49,
|
| 1057 |
+
"learning_rate": 3.782676442354513e-05,
|
| 1058 |
+
"loss": 2.3821,
|
| 1059 |
+
"step": 35000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.49,
|
| 1063 |
+
"learning_rate": 3.7757203077393956e-05,
|
| 1064 |
+
"loss": 2.4588,
|
| 1065 |
+
"step": 35200
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.49,
|
| 1069 |
+
"learning_rate": 3.7687641731242785e-05,
|
| 1070 |
+
"loss": 2.4303,
|
| 1071 |
+
"step": 35400
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.5,
|
| 1075 |
+
"learning_rate": 3.7618080385091614e-05,
|
| 1076 |
+
"loss": 2.4043,
|
| 1077 |
+
"step": 35600
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 0.5,
|
| 1081 |
+
"learning_rate": 3.754851903894044e-05,
|
| 1082 |
+
"loss": 2.3885,
|
| 1083 |
+
"step": 35800
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 0.5,
|
| 1087 |
+
"learning_rate": 3.747895769278927e-05,
|
| 1088 |
+
"loss": 2.452,
|
| 1089 |
+
"step": 36000
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.5,
|
| 1093 |
+
"learning_rate": 3.740939634663811e-05,
|
| 1094 |
+
"loss": 2.4303,
|
| 1095 |
+
"step": 36200
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.51,
|
| 1099 |
+
"learning_rate": 3.7339835000486936e-05,
|
| 1100 |
+
"loss": 2.4219,
|
| 1101 |
+
"step": 36400
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.51,
|
| 1105 |
+
"learning_rate": 3.727027365433576e-05,
|
| 1106 |
+
"loss": 2.4514,
|
| 1107 |
+
"step": 36600
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.51,
|
| 1111 |
+
"learning_rate": 3.720071230818459e-05,
|
| 1112 |
+
"loss": 2.3961,
|
| 1113 |
+
"step": 36800
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 0.51,
|
| 1117 |
+
"learning_rate": 3.7131150962033416e-05,
|
| 1118 |
+
"loss": 2.3741,
|
| 1119 |
+
"step": 37000
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.52,
|
| 1123 |
+
"learning_rate": 3.7061589615882244e-05,
|
| 1124 |
+
"loss": 2.4553,
|
| 1125 |
+
"step": 37200
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.52,
|
| 1129 |
+
"learning_rate": 3.699202826973108e-05,
|
| 1130 |
+
"loss": 2.3806,
|
| 1131 |
+
"step": 37400
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.52,
|
| 1135 |
+
"learning_rate": 3.692246692357991e-05,
|
| 1136 |
+
"loss": 2.416,
|
| 1137 |
+
"step": 37600
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.53,
|
| 1141 |
+
"learning_rate": 3.685290557742874e-05,
|
| 1142 |
+
"loss": 2.3745,
|
| 1143 |
+
"step": 37800
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.53,
|
| 1147 |
+
"learning_rate": 3.678334423127757e-05,
|
| 1148 |
+
"loss": 2.4195,
|
| 1149 |
+
"step": 38000
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.53,
|
| 1153 |
+
"learning_rate": 3.6713782885126395e-05,
|
| 1154 |
+
"loss": 2.4216,
|
| 1155 |
+
"step": 38200
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.53,
|
| 1159 |
+
"learning_rate": 3.6644221538975224e-05,
|
| 1160 |
+
"loss": 2.3769,
|
| 1161 |
+
"step": 38400
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.54,
|
| 1165 |
+
"learning_rate": 3.657466019282405e-05,
|
| 1166 |
+
"loss": 2.3851,
|
| 1167 |
+
"step": 38600
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.54,
|
| 1171 |
+
"learning_rate": 3.650509884667288e-05,
|
| 1172 |
+
"loss": 2.4112,
|
| 1173 |
+
"step": 38800
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.54,
|
| 1177 |
+
"learning_rate": 3.643553750052171e-05,
|
| 1178 |
+
"loss": 2.4096,
|
| 1179 |
+
"step": 39000
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.55,
|
| 1183 |
+
"learning_rate": 3.636597615437054e-05,
|
| 1184 |
+
"loss": 2.4354,
|
| 1185 |
+
"step": 39200
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.55,
|
| 1189 |
+
"learning_rate": 3.629641480821937e-05,
|
| 1190 |
+
"loss": 2.422,
|
| 1191 |
+
"step": 39400
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.55,
|
| 1195 |
+
"learning_rate": 3.62268534620682e-05,
|
| 1196 |
+
"loss": 2.4195,
|
| 1197 |
+
"step": 39600
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.55,
|
| 1201 |
+
"learning_rate": 3.6157292115917026e-05,
|
| 1202 |
+
"loss": 2.4506,
|
| 1203 |
+
"step": 39800
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 0.56,
|
| 1207 |
+
"learning_rate": 3.608773076976586e-05,
|
| 1208 |
+
"loss": 2.4016,
|
| 1209 |
+
"step": 40000
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.56,
|
| 1213 |
+
"learning_rate": 3.601816942361469e-05,
|
| 1214 |
+
"loss": 2.3905,
|
| 1215 |
+
"step": 40200
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.56,
|
| 1219 |
+
"learning_rate": 3.594860807746352e-05,
|
| 1220 |
+
"loss": 2.4293,
|
| 1221 |
+
"step": 40400
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.56,
|
| 1225 |
+
"learning_rate": 3.587904673131234e-05,
|
| 1226 |
+
"loss": 2.4237,
|
| 1227 |
+
"step": 40600
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.57,
|
| 1231 |
+
"learning_rate": 3.580948538516117e-05,
|
| 1232 |
+
"loss": 2.3867,
|
| 1233 |
+
"step": 40800
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.57,
|
| 1237 |
+
"learning_rate": 3.5739924039010006e-05,
|
| 1238 |
+
"loss": 2.3885,
|
| 1239 |
+
"step": 41000
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.57,
|
| 1243 |
+
"learning_rate": 3.5670362692858835e-05,
|
| 1244 |
+
"loss": 2.4321,
|
| 1245 |
+
"step": 41200
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.58,
|
| 1249 |
+
"learning_rate": 3.5600801346707664e-05,
|
| 1250 |
+
"loss": 2.3701,
|
| 1251 |
+
"step": 41400
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.58,
|
| 1255 |
+
"learning_rate": 3.553124000055649e-05,
|
| 1256 |
+
"loss": 2.4581,
|
| 1257 |
+
"step": 41600
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.58,
|
| 1261 |
+
"learning_rate": 3.546167865440532e-05,
|
| 1262 |
+
"loss": 2.398,
|
| 1263 |
+
"step": 41800
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.58,
|
| 1267 |
+
"learning_rate": 3.539211730825415e-05,
|
| 1268 |
+
"loss": 2.4155,
|
| 1269 |
+
"step": 42000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.59,
|
| 1273 |
+
"learning_rate": 3.532255596210298e-05,
|
| 1274 |
+
"loss": 2.4122,
|
| 1275 |
+
"step": 42200
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 0.59,
|
| 1279 |
+
"learning_rate": 3.5252994615951815e-05,
|
| 1280 |
+
"loss": 2.3833,
|
| 1281 |
+
"step": 42400
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 0.59,
|
| 1285 |
+
"learning_rate": 3.518343326980064e-05,
|
| 1286 |
+
"loss": 2.4614,
|
| 1287 |
+
"step": 42600
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 0.6,
|
| 1291 |
+
"learning_rate": 3.5113871923649466e-05,
|
| 1292 |
+
"loss": 2.3859,
|
| 1293 |
+
"step": 42800
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 0.6,
|
| 1297 |
+
"learning_rate": 3.5044310577498295e-05,
|
| 1298 |
+
"loss": 2.3674,
|
| 1299 |
+
"step": 43000
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.6,
|
| 1303 |
+
"learning_rate": 3.4974749231347124e-05,
|
| 1304 |
+
"loss": 2.4227,
|
| 1305 |
+
"step": 43200
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.6,
|
| 1309 |
+
"learning_rate": 3.490518788519595e-05,
|
| 1310 |
+
"loss": 2.4031,
|
| 1311 |
+
"step": 43400
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.61,
|
| 1315 |
+
"learning_rate": 3.483562653904479e-05,
|
| 1316 |
+
"loss": 2.4271,
|
| 1317 |
+
"step": 43600
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 0.61,
|
| 1321 |
+
"learning_rate": 3.476606519289362e-05,
|
| 1322 |
+
"loss": 2.4176,
|
| 1323 |
+
"step": 43800
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 0.61,
|
| 1327 |
+
"learning_rate": 3.4696503846742446e-05,
|
| 1328 |
+
"loss": 2.4076,
|
| 1329 |
+
"step": 44000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 0.61,
|
| 1333 |
+
"learning_rate": 3.4626942500591275e-05,
|
| 1334 |
+
"loss": 2.3618,
|
| 1335 |
+
"step": 44200
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 0.62,
|
| 1339 |
+
"learning_rate": 3.4557381154440104e-05,
|
| 1340 |
+
"loss": 2.3998,
|
| 1341 |
+
"step": 44400
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.62,
|
| 1345 |
+
"learning_rate": 3.448781980828893e-05,
|
| 1346 |
+
"loss": 2.4145,
|
| 1347 |
+
"step": 44600
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.62,
|
| 1351 |
+
"learning_rate": 3.441825846213776e-05,
|
| 1352 |
+
"loss": 2.3621,
|
| 1353 |
+
"step": 44800
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.63,
|
| 1357 |
+
"learning_rate": 3.434869711598659e-05,
|
| 1358 |
+
"loss": 2.4247,
|
| 1359 |
+
"step": 45000
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 0.63,
|
| 1363 |
+
"learning_rate": 3.427913576983542e-05,
|
| 1364 |
+
"loss": 2.3934,
|
| 1365 |
+
"step": 45200
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 0.63,
|
| 1369 |
+
"learning_rate": 3.420957442368425e-05,
|
| 1370 |
+
"loss": 2.379,
|
| 1371 |
+
"step": 45400
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 0.63,
|
| 1375 |
+
"learning_rate": 3.414001307753308e-05,
|
| 1376 |
+
"loss": 2.3932,
|
| 1377 |
+
"step": 45600
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 0.64,
|
| 1381 |
+
"learning_rate": 3.4070451731381906e-05,
|
| 1382 |
+
"loss": 2.3922,
|
| 1383 |
+
"step": 45800
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 0.64,
|
| 1387 |
+
"learning_rate": 3.400089038523074e-05,
|
| 1388 |
+
"loss": 2.4221,
|
| 1389 |
+
"step": 46000
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.64,
|
| 1393 |
+
"learning_rate": 3.393132903907957e-05,
|
| 1394 |
+
"loss": 2.3921,
|
| 1395 |
+
"step": 46200
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.65,
|
| 1399 |
+
"learning_rate": 3.38617676929284e-05,
|
| 1400 |
+
"loss": 2.3484,
|
| 1401 |
+
"step": 46400
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 0.65,
|
| 1405 |
+
"learning_rate": 3.379220634677722e-05,
|
| 1406 |
+
"loss": 2.3812,
|
| 1407 |
+
"step": 46600
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 0.65,
|
| 1411 |
+
"learning_rate": 3.372264500062605e-05,
|
| 1412 |
+
"loss": 2.3616,
|
| 1413 |
+
"step": 46800
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 0.65,
|
| 1417 |
+
"learning_rate": 3.365308365447488e-05,
|
| 1418 |
+
"loss": 2.388,
|
| 1419 |
+
"step": 47000
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.66,
|
| 1423 |
+
"learning_rate": 3.3583522308323714e-05,
|
| 1424 |
+
"loss": 2.3102,
|
| 1425 |
+
"step": 47200
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 0.66,
|
| 1429 |
+
"learning_rate": 3.351396096217254e-05,
|
| 1430 |
+
"loss": 2.3746,
|
| 1431 |
+
"step": 47400
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.66,
|
| 1435 |
+
"learning_rate": 3.344439961602137e-05,
|
| 1436 |
+
"loss": 2.4212,
|
| 1437 |
+
"step": 47600
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.67,
|
| 1441 |
+
"learning_rate": 3.33748382698702e-05,
|
| 1442 |
+
"loss": 2.3558,
|
| 1443 |
+
"step": 47800
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 0.67,
|
| 1447 |
+
"learning_rate": 3.330527692371903e-05,
|
| 1448 |
+
"loss": 2.393,
|
| 1449 |
+
"step": 48000
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.67,
|
| 1453 |
+
"learning_rate": 3.323571557756786e-05,
|
| 1454 |
+
"loss": 2.3798,
|
| 1455 |
+
"step": 48200
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 0.67,
|
| 1459 |
+
"learning_rate": 3.316615423141669e-05,
|
| 1460 |
+
"loss": 2.4259,
|
| 1461 |
+
"step": 48400
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 0.68,
|
| 1465 |
+
"learning_rate": 3.3096592885265516e-05,
|
| 1466 |
+
"loss": 2.3816,
|
| 1467 |
+
"step": 48600
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 0.68,
|
| 1471 |
+
"learning_rate": 3.3027031539114345e-05,
|
| 1472 |
+
"loss": 2.3699,
|
| 1473 |
+
"step": 48800
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.68,
|
| 1477 |
+
"learning_rate": 3.2957470192963174e-05,
|
| 1478 |
+
"loss": 2.4018,
|
| 1479 |
+
"step": 49000
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.68,
|
| 1483 |
+
"learning_rate": 3.2887908846812e-05,
|
| 1484 |
+
"loss": 2.3356,
|
| 1485 |
+
"step": 49200
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 0.69,
|
| 1489 |
+
"learning_rate": 3.281834750066083e-05,
|
| 1490 |
+
"loss": 2.347,
|
| 1491 |
+
"step": 49400
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 0.69,
|
| 1495 |
+
"learning_rate": 3.274878615450967e-05,
|
| 1496 |
+
"loss": 2.3598,
|
| 1497 |
+
"step": 49600
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 0.69,
|
| 1501 |
+
"learning_rate": 3.2679224808358496e-05,
|
| 1502 |
+
"loss": 2.3773,
|
| 1503 |
+
"step": 49800
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 0.7,
|
| 1507 |
+
"learning_rate": 3.2609663462207325e-05,
|
| 1508 |
+
"loss": 2.3704,
|
| 1509 |
+
"step": 50000
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 0.7,
|
| 1513 |
+
"learning_rate": 3.2540102116056154e-05,
|
| 1514 |
+
"loss": 2.3317,
|
| 1515 |
+
"step": 50200
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.7,
|
| 1519 |
+
"learning_rate": 3.247054076990498e-05,
|
| 1520 |
+
"loss": 2.3878,
|
| 1521 |
+
"step": 50400
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.7,
|
| 1525 |
+
"learning_rate": 3.2400979423753805e-05,
|
| 1526 |
+
"loss": 2.3198,
|
| 1527 |
+
"step": 50600
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 0.71,
|
| 1531 |
+
"learning_rate": 3.233141807760264e-05,
|
| 1532 |
+
"loss": 2.4416,
|
| 1533 |
+
"step": 50800
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 0.71,
|
| 1537 |
+
"learning_rate": 3.226185673145147e-05,
|
| 1538 |
+
"loss": 2.3275,
|
| 1539 |
+
"step": 51000
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 0.71,
|
| 1543 |
+
"learning_rate": 3.21922953853003e-05,
|
| 1544 |
+
"loss": 2.3858,
|
| 1545 |
+
"step": 51200
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 0.72,
|
| 1549 |
+
"learning_rate": 3.212273403914913e-05,
|
| 1550 |
+
"loss": 2.3448,
|
| 1551 |
+
"step": 51400
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 0.72,
|
| 1555 |
+
"learning_rate": 3.2053172692997956e-05,
|
| 1556 |
+
"loss": 2.3493,
|
| 1557 |
+
"step": 51600
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.72,
|
| 1561 |
+
"learning_rate": 3.1983611346846785e-05,
|
| 1562 |
+
"loss": 2.3644,
|
| 1563 |
+
"step": 51800
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.72,
|
| 1567 |
+
"learning_rate": 3.1914050000695614e-05,
|
| 1568 |
+
"loss": 2.339,
|
| 1569 |
+
"step": 52000
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 0.73,
|
| 1573 |
+
"learning_rate": 3.184448865454445e-05,
|
| 1574 |
+
"loss": 2.3393,
|
| 1575 |
+
"step": 52200
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 0.73,
|
| 1579 |
+
"learning_rate": 3.177492730839327e-05,
|
| 1580 |
+
"loss": 2.3754,
|
| 1581 |
+
"step": 52400
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 0.73,
|
| 1585 |
+
"learning_rate": 3.17053659622421e-05,
|
| 1586 |
+
"loss": 2.3251,
|
| 1587 |
+
"step": 52600
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 0.73,
|
| 1591 |
+
"learning_rate": 3.163580461609093e-05,
|
| 1592 |
+
"loss": 2.3407,
|
| 1593 |
+
"step": 52800
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.74,
|
| 1597 |
+
"learning_rate": 3.156624326993976e-05,
|
| 1598 |
+
"loss": 2.3092,
|
| 1599 |
+
"step": 53000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.74,
|
| 1603 |
+
"learning_rate": 3.149668192378859e-05,
|
| 1604 |
+
"loss": 2.3626,
|
| 1605 |
+
"step": 53200
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.74,
|
| 1609 |
+
"learning_rate": 3.142712057763742e-05,
|
| 1610 |
+
"loss": 2.3639,
|
| 1611 |
+
"step": 53400
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.75,
|
| 1615 |
+
"learning_rate": 3.135755923148625e-05,
|
| 1616 |
+
"loss": 2.3443,
|
| 1617 |
+
"step": 53600
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 0.75,
|
| 1621 |
+
"learning_rate": 3.128799788533508e-05,
|
| 1622 |
+
"loss": 2.3274,
|
| 1623 |
+
"step": 53800
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 0.75,
|
| 1627 |
+
"learning_rate": 3.121843653918391e-05,
|
| 1628 |
+
"loss": 2.3728,
|
| 1629 |
+
"step": 54000
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.75,
|
| 1633 |
+
"learning_rate": 3.114887519303274e-05,
|
| 1634 |
+
"loss": 2.3047,
|
| 1635 |
+
"step": 54200
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 0.76,
|
| 1639 |
+
"learning_rate": 3.107931384688157e-05,
|
| 1640 |
+
"loss": 2.316,
|
| 1641 |
+
"step": 54400
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.76,
|
| 1645 |
+
"learning_rate": 3.1009752500730396e-05,
|
| 1646 |
+
"loss": 2.3326,
|
| 1647 |
+
"step": 54600
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.76,
|
| 1651 |
+
"learning_rate": 3.0940191154579225e-05,
|
| 1652 |
+
"loss": 2.357,
|
| 1653 |
+
"step": 54800
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.77,
|
| 1657 |
+
"learning_rate": 3.0870629808428053e-05,
|
| 1658 |
+
"loss": 2.367,
|
| 1659 |
+
"step": 55000
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 0.77,
|
| 1663 |
+
"learning_rate": 3.080106846227688e-05,
|
| 1664 |
+
"loss": 2.3529,
|
| 1665 |
+
"step": 55200
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 0.77,
|
| 1669 |
+
"learning_rate": 3.073150711612571e-05,
|
| 1670 |
+
"loss": 2.3748,
|
| 1671 |
+
"step": 55400
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 0.77,
|
| 1675 |
+
"learning_rate": 3.066194576997454e-05,
|
| 1676 |
+
"loss": 2.3208,
|
| 1677 |
+
"step": 55600
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 0.78,
|
| 1681 |
+
"learning_rate": 3.0592384423823376e-05,
|
| 1682 |
+
"loss": 2.3327,
|
| 1683 |
+
"step": 55800
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.78,
|
| 1687 |
+
"learning_rate": 3.0522823077672204e-05,
|
| 1688 |
+
"loss": 2.3327,
|
| 1689 |
+
"step": 56000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.78,
|
| 1693 |
+
"learning_rate": 3.0453261731521033e-05,
|
| 1694 |
+
"loss": 2.3588,
|
| 1695 |
+
"step": 56200
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 0.78,
|
| 1699 |
+
"learning_rate": 3.0383700385369855e-05,
|
| 1700 |
+
"loss": 2.3343,
|
| 1701 |
+
"step": 56400
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 0.79,
|
| 1705 |
+
"learning_rate": 3.0314139039218688e-05,
|
| 1706 |
+
"loss": 2.3095,
|
| 1707 |
+
"step": 56600
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 0.79,
|
| 1711 |
+
"learning_rate": 3.0244577693067517e-05,
|
| 1712 |
+
"loss": 2.3547,
|
| 1713 |
+
"step": 56800
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 0.79,
|
| 1717 |
+
"learning_rate": 3.0175016346916345e-05,
|
| 1718 |
+
"loss": 2.3346,
|
| 1719 |
+
"step": 57000
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 0.8,
|
| 1723 |
+
"learning_rate": 3.0105455000765174e-05,
|
| 1724 |
+
"loss": 2.289,
|
| 1725 |
+
"step": 57200
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.8,
|
| 1729 |
+
"learning_rate": 3.0035893654614006e-05,
|
| 1730 |
+
"loss": 2.3577,
|
| 1731 |
+
"step": 57400
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.8,
|
| 1735 |
+
"learning_rate": 2.9966332308462835e-05,
|
| 1736 |
+
"loss": 2.3592,
|
| 1737 |
+
"step": 57600
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 0.8,
|
| 1741 |
+
"learning_rate": 2.9896770962311664e-05,
|
| 1742 |
+
"loss": 2.3258,
|
| 1743 |
+
"step": 57800
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 0.81,
|
| 1747 |
+
"learning_rate": 2.9827209616160496e-05,
|
| 1748 |
+
"loss": 2.336,
|
| 1749 |
+
"step": 58000
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 0.81,
|
| 1753 |
+
"learning_rate": 2.9757648270009325e-05,
|
| 1754 |
+
"loss": 2.3663,
|
| 1755 |
+
"step": 58200
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 0.81,
|
| 1759 |
+
"learning_rate": 2.968808692385815e-05,
|
| 1760 |
+
"loss": 2.3952,
|
| 1761 |
+
"step": 58400
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 0.82,
|
| 1765 |
+
"learning_rate": 2.961852557770698e-05,
|
| 1766 |
+
"loss": 2.3366,
|
| 1767 |
+
"step": 58600
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.82,
|
| 1771 |
+
"learning_rate": 2.954896423155581e-05,
|
| 1772 |
+
"loss": 2.316,
|
| 1773 |
+
"step": 58800
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.82,
|
| 1777 |
+
"learning_rate": 2.9479402885404637e-05,
|
| 1778 |
+
"loss": 2.3167,
|
| 1779 |
+
"step": 59000
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 0.82,
|
| 1783 |
+
"learning_rate": 2.940984153925347e-05,
|
| 1784 |
+
"loss": 2.3853,
|
| 1785 |
+
"step": 59200
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 0.83,
|
| 1789 |
+
"learning_rate": 2.93402801931023e-05,
|
| 1790 |
+
"loss": 2.3824,
|
| 1791 |
+
"step": 59400
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 0.83,
|
| 1795 |
+
"learning_rate": 2.9270718846951127e-05,
|
| 1796 |
+
"loss": 2.385,
|
| 1797 |
+
"step": 59600
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 0.83,
|
| 1801 |
+
"learning_rate": 2.920115750079996e-05,
|
| 1802 |
+
"loss": 2.3916,
|
| 1803 |
+
"step": 59800
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 0.83,
|
| 1807 |
+
"learning_rate": 2.913159615464879e-05,
|
| 1808 |
+
"loss": 2.3757,
|
| 1809 |
+
"step": 60000
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.84,
|
| 1813 |
+
"learning_rate": 2.9062034808497617e-05,
|
| 1814 |
+
"loss": 2.3031,
|
| 1815 |
+
"step": 60200
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.84,
|
| 1819 |
+
"learning_rate": 2.8992473462346443e-05,
|
| 1820 |
+
"loss": 2.3391,
|
| 1821 |
+
"step": 60400
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 0.84,
|
| 1825 |
+
"learning_rate": 2.892291211619527e-05,
|
| 1826 |
+
"loss": 2.3327,
|
| 1827 |
+
"step": 60600
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 0.85,
|
| 1831 |
+
"learning_rate": 2.88533507700441e-05,
|
| 1832 |
+
"loss": 2.3553,
|
| 1833 |
+
"step": 60800
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 0.85,
|
| 1837 |
+
"learning_rate": 2.8783789423892933e-05,
|
| 1838 |
+
"loss": 2.4071,
|
| 1839 |
+
"step": 61000
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 0.85,
|
| 1843 |
+
"learning_rate": 2.871422807774176e-05,
|
| 1844 |
+
"loss": 2.2922,
|
| 1845 |
+
"step": 61200
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 0.85,
|
| 1849 |
+
"learning_rate": 2.864466673159059e-05,
|
| 1850 |
+
"loss": 2.2956,
|
| 1851 |
+
"step": 61400
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 0.86,
|
| 1855 |
+
"learning_rate": 2.8575105385439423e-05,
|
| 1856 |
+
"loss": 2.3518,
|
| 1857 |
+
"step": 61600
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 0.86,
|
| 1861 |
+
"learning_rate": 2.850554403928825e-05,
|
| 1862 |
+
"loss": 2.3378,
|
| 1863 |
+
"step": 61800
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 0.86,
|
| 1867 |
+
"learning_rate": 2.843598269313708e-05,
|
| 1868 |
+
"loss": 2.3567,
|
| 1869 |
+
"step": 62000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 0.87,
|
| 1873 |
+
"learning_rate": 2.836642134698591e-05,
|
| 1874 |
+
"loss": 2.3835,
|
| 1875 |
+
"step": 62200
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 0.87,
|
| 1879 |
+
"learning_rate": 2.8296860000834735e-05,
|
| 1880 |
+
"loss": 2.3527,
|
| 1881 |
+
"step": 62400
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 0.87,
|
| 1885 |
+
"learning_rate": 2.8227298654683564e-05,
|
| 1886 |
+
"loss": 2.3609,
|
| 1887 |
+
"step": 62600
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 0.87,
|
| 1891 |
+
"learning_rate": 2.8157737308532396e-05,
|
| 1892 |
+
"loss": 2.3263,
|
| 1893 |
+
"step": 62800
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 0.88,
|
| 1897 |
+
"learning_rate": 2.8088175962381225e-05,
|
| 1898 |
+
"loss": 2.3604,
|
| 1899 |
+
"step": 63000
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.88,
|
| 1903 |
+
"learning_rate": 2.8018614616230054e-05,
|
| 1904 |
+
"loss": 2.371,
|
| 1905 |
+
"step": 63200
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 0.88,
|
| 1909 |
+
"learning_rate": 2.7949053270078886e-05,
|
| 1910 |
+
"loss": 2.3555,
|
| 1911 |
+
"step": 63400
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 0.88,
|
| 1915 |
+
"learning_rate": 2.7879491923927715e-05,
|
| 1916 |
+
"loss": 2.3573,
|
| 1917 |
+
"step": 63600
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 0.89,
|
| 1921 |
+
"learning_rate": 2.7809930577776543e-05,
|
| 1922 |
+
"loss": 2.3151,
|
| 1923 |
+
"step": 63800
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 0.89,
|
| 1927 |
+
"learning_rate": 2.7740369231625372e-05,
|
| 1928 |
+
"loss": 2.342,
|
| 1929 |
+
"step": 64000
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 0.89,
|
| 1933 |
+
"learning_rate": 2.7670807885474205e-05,
|
| 1934 |
+
"loss": 2.3264,
|
| 1935 |
+
"step": 64200
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.9,
|
| 1939 |
+
"learning_rate": 2.7601246539323027e-05,
|
| 1940 |
+
"loss": 2.3124,
|
| 1941 |
+
"step": 64400
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.9,
|
| 1945 |
+
"learning_rate": 2.753168519317186e-05,
|
| 1946 |
+
"loss": 2.3672,
|
| 1947 |
+
"step": 64600
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 0.9,
|
| 1951 |
+
"learning_rate": 2.7462123847020688e-05,
|
| 1952 |
+
"loss": 2.3758,
|
| 1953 |
+
"step": 64800
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 0.9,
|
| 1957 |
+
"learning_rate": 2.7392562500869517e-05,
|
| 1958 |
+
"loss": 2.3426,
|
| 1959 |
+
"step": 65000
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 0.91,
|
| 1963 |
+
"learning_rate": 2.732300115471835e-05,
|
| 1964 |
+
"loss": 2.3127,
|
| 1965 |
+
"step": 65200
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 0.91,
|
| 1969 |
+
"learning_rate": 2.7253439808567178e-05,
|
| 1970 |
+
"loss": 2.351,
|
| 1971 |
+
"step": 65400
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 0.91,
|
| 1975 |
+
"learning_rate": 2.7183878462416007e-05,
|
| 1976 |
+
"loss": 2.3214,
|
| 1977 |
+
"step": 65600
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 0.92,
|
| 1981 |
+
"learning_rate": 2.7114317116264835e-05,
|
| 1982 |
+
"loss": 2.3141,
|
| 1983 |
+
"step": 65800
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 0.92,
|
| 1987 |
+
"learning_rate": 2.7044755770113668e-05,
|
| 1988 |
+
"loss": 2.3095,
|
| 1989 |
+
"step": 66000
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 0.92,
|
| 1993 |
+
"learning_rate": 2.6975194423962497e-05,
|
| 1994 |
+
"loss": 2.3303,
|
| 1995 |
+
"step": 66200
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 0.92,
|
| 1999 |
+
"learning_rate": 2.6905633077811322e-05,
|
| 2000 |
+
"loss": 2.3111,
|
| 2001 |
+
"step": 66400
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 0.93,
|
| 2005 |
+
"learning_rate": 2.683607173166015e-05,
|
| 2006 |
+
"loss": 2.2752,
|
| 2007 |
+
"step": 66600
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 0.93,
|
| 2011 |
+
"learning_rate": 2.676651038550898e-05,
|
| 2012 |
+
"loss": 2.3245,
|
| 2013 |
+
"step": 66800
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 0.93,
|
| 2017 |
+
"learning_rate": 2.6696949039357812e-05,
|
| 2018 |
+
"loss": 2.3506,
|
| 2019 |
+
"step": 67000
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 0.93,
|
| 2023 |
+
"learning_rate": 2.662738769320664e-05,
|
| 2024 |
+
"loss": 2.2878,
|
| 2025 |
+
"step": 67200
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 0.94,
|
| 2029 |
+
"learning_rate": 2.655782634705547e-05,
|
| 2030 |
+
"loss": 2.2858,
|
| 2031 |
+
"step": 67400
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 0.94,
|
| 2035 |
+
"learning_rate": 2.64882650009043e-05,
|
| 2036 |
+
"loss": 2.3803,
|
| 2037 |
+
"step": 67600
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 0.94,
|
| 2041 |
+
"learning_rate": 2.641870365475313e-05,
|
| 2042 |
+
"loss": 2.334,
|
| 2043 |
+
"step": 67800
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 0.95,
|
| 2047 |
+
"learning_rate": 2.634914230860196e-05,
|
| 2048 |
+
"loss": 2.3088,
|
| 2049 |
+
"step": 68000
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 0.95,
|
| 2053 |
+
"learning_rate": 2.627958096245079e-05,
|
| 2054 |
+
"loss": 2.3356,
|
| 2055 |
+
"step": 68200
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 0.95,
|
| 2059 |
+
"learning_rate": 2.6210019616299614e-05,
|
| 2060 |
+
"loss": 2.321,
|
| 2061 |
+
"step": 68400
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 0.95,
|
| 2065 |
+
"learning_rate": 2.6140458270148443e-05,
|
| 2066 |
+
"loss": 2.3396,
|
| 2067 |
+
"step": 68600
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 0.96,
|
| 2071 |
+
"learning_rate": 2.607089692399727e-05,
|
| 2072 |
+
"loss": 2.2976,
|
| 2073 |
+
"step": 68800
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 0.96,
|
| 2077 |
+
"learning_rate": 2.6001335577846104e-05,
|
| 2078 |
+
"loss": 2.2884,
|
| 2079 |
+
"step": 69000
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 0.96,
|
| 2083 |
+
"learning_rate": 2.5931774231694933e-05,
|
| 2084 |
+
"loss": 2.3099,
|
| 2085 |
+
"step": 69200
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 0.97,
|
| 2089 |
+
"learning_rate": 2.586221288554376e-05,
|
| 2090 |
+
"loss": 2.3217,
|
| 2091 |
+
"step": 69400
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 0.97,
|
| 2095 |
+
"learning_rate": 2.5792651539392594e-05,
|
| 2096 |
+
"loss": 2.2856,
|
| 2097 |
+
"step": 69600
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 0.97,
|
| 2101 |
+
"learning_rate": 2.5723090193241423e-05,
|
| 2102 |
+
"loss": 2.3615,
|
| 2103 |
+
"step": 69800
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 0.97,
|
| 2107 |
+
"learning_rate": 2.565352884709025e-05,
|
| 2108 |
+
"loss": 2.2836,
|
| 2109 |
+
"step": 70000
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 0.98,
|
| 2113 |
+
"learning_rate": 2.5583967500939077e-05,
|
| 2114 |
+
"loss": 2.3666,
|
| 2115 |
+
"step": 70200
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 0.98,
|
| 2119 |
+
"learning_rate": 2.5514406154787906e-05,
|
| 2120 |
+
"loss": 2.3123,
|
| 2121 |
+
"step": 70400
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 0.98,
|
| 2125 |
+
"learning_rate": 2.5444844808636735e-05,
|
| 2126 |
+
"loss": 2.3703,
|
| 2127 |
+
"step": 70600
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 0.98,
|
| 2131 |
+
"learning_rate": 2.5375283462485567e-05,
|
| 2132 |
+
"loss": 2.3,
|
| 2133 |
+
"step": 70800
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 0.99,
|
| 2137 |
+
"learning_rate": 2.5305722116334396e-05,
|
| 2138 |
+
"loss": 2.3252,
|
| 2139 |
+
"step": 71000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 0.99,
|
| 2143 |
+
"learning_rate": 2.5236160770183225e-05,
|
| 2144 |
+
"loss": 2.2764,
|
| 2145 |
+
"step": 71200
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 0.99,
|
| 2149 |
+
"learning_rate": 2.5166599424032057e-05,
|
| 2150 |
+
"loss": 2.3298,
|
| 2151 |
+
"step": 71400
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 1.0,
|
| 2155 |
+
"learning_rate": 2.5097038077880886e-05,
|
| 2156 |
+
"loss": 2.2929,
|
| 2157 |
+
"step": 71600
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 1.0,
|
| 2161 |
+
"learning_rate": 2.5027476731729715e-05,
|
| 2162 |
+
"loss": 2.2647,
|
| 2163 |
+
"step": 71800
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 1.0,
|
| 2167 |
+
"learning_rate": 2.4957915385578544e-05,
|
| 2168 |
+
"loss": 2.2967,
|
| 2169 |
+
"step": 72000
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 1.0,
|
| 2173 |
+
"learning_rate": 2.4888354039427372e-05,
|
| 2174 |
+
"loss": 2.2854,
|
| 2175 |
+
"step": 72200
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 1.01,
|
| 2179 |
+
"learning_rate": 2.48187926932762e-05,
|
| 2180 |
+
"loss": 2.2653,
|
| 2181 |
+
"step": 72400
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 1.01,
|
| 2185 |
+
"learning_rate": 2.4749231347125034e-05,
|
| 2186 |
+
"loss": 2.2476,
|
| 2187 |
+
"step": 72600
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 1.01,
|
| 2191 |
+
"learning_rate": 2.467967000097386e-05,
|
| 2192 |
+
"loss": 2.2801,
|
| 2193 |
+
"step": 72800
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 1.02,
|
| 2197 |
+
"learning_rate": 2.4610108654822688e-05,
|
| 2198 |
+
"loss": 2.2836,
|
| 2199 |
+
"step": 73000
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 1.02,
|
| 2203 |
+
"learning_rate": 2.454054730867152e-05,
|
| 2204 |
+
"loss": 2.2847,
|
| 2205 |
+
"step": 73200
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 1.02,
|
| 2209 |
+
"learning_rate": 2.447098596252035e-05,
|
| 2210 |
+
"loss": 2.3052,
|
| 2211 |
+
"step": 73400
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 1.02,
|
| 2215 |
+
"learning_rate": 2.4401424616369178e-05,
|
| 2216 |
+
"loss": 2.285,
|
| 2217 |
+
"step": 73600
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 1.03,
|
| 2221 |
+
"learning_rate": 2.4331863270218007e-05,
|
| 2222 |
+
"loss": 2.22,
|
| 2223 |
+
"step": 73800
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 1.03,
|
| 2227 |
+
"learning_rate": 2.4262301924066836e-05,
|
| 2228 |
+
"loss": 2.2589,
|
| 2229 |
+
"step": 74000
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 1.03,
|
| 2233 |
+
"learning_rate": 2.4192740577915664e-05,
|
| 2234 |
+
"loss": 2.2705,
|
| 2235 |
+
"step": 74200
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 1.04,
|
| 2239 |
+
"learning_rate": 2.4123179231764497e-05,
|
| 2240 |
+
"loss": 2.304,
|
| 2241 |
+
"step": 74400
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 1.04,
|
| 2245 |
+
"learning_rate": 2.4053617885613325e-05,
|
| 2246 |
+
"loss": 2.2499,
|
| 2247 |
+
"step": 74600
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 1.04,
|
| 2251 |
+
"learning_rate": 2.398405653946215e-05,
|
| 2252 |
+
"loss": 2.3109,
|
| 2253 |
+
"step": 74800
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 1.04,
|
| 2257 |
+
"learning_rate": 2.3914495193310983e-05,
|
| 2258 |
+
"loss": 2.2772,
|
| 2259 |
+
"step": 75000
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 1.05,
|
| 2263 |
+
"learning_rate": 2.3844933847159812e-05,
|
| 2264 |
+
"loss": 2.2835,
|
| 2265 |
+
"step": 75200
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 1.05,
|
| 2269 |
+
"learning_rate": 2.377537250100864e-05,
|
| 2270 |
+
"loss": 2.298,
|
| 2271 |
+
"step": 75400
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 1.05,
|
| 2275 |
+
"learning_rate": 2.370581115485747e-05,
|
| 2276 |
+
"loss": 2.2486,
|
| 2277 |
+
"step": 75600
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 1.05,
|
| 2281 |
+
"learning_rate": 2.36362498087063e-05,
|
| 2282 |
+
"loss": 2.304,
|
| 2283 |
+
"step": 75800
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 1.06,
|
| 2287 |
+
"learning_rate": 2.3566688462555127e-05,
|
| 2288 |
+
"loss": 2.2604,
|
| 2289 |
+
"step": 76000
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 1.06,
|
| 2293 |
+
"learning_rate": 2.3497127116403956e-05,
|
| 2294 |
+
"loss": 2.2368,
|
| 2295 |
+
"step": 76200
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 1.06,
|
| 2299 |
+
"learning_rate": 2.342756577025279e-05,
|
| 2300 |
+
"loss": 2.2736,
|
| 2301 |
+
"step": 76400
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 1.07,
|
| 2305 |
+
"learning_rate": 2.3358004424101617e-05,
|
| 2306 |
+
"loss": 2.3323,
|
| 2307 |
+
"step": 76600
|
| 2308 |
+
},
|
| 2309 |
+
{
|
| 2310 |
+
"epoch": 1.07,
|
| 2311 |
+
"learning_rate": 2.3288443077950446e-05,
|
| 2312 |
+
"loss": 2.2373,
|
| 2313 |
+
"step": 76800
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 1.07,
|
| 2317 |
+
"learning_rate": 2.3218881731799275e-05,
|
| 2318 |
+
"loss": 2.3072,
|
| 2319 |
+
"step": 77000
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 1.07,
|
| 2323 |
+
"learning_rate": 2.3149320385648104e-05,
|
| 2324 |
+
"loss": 2.2919,
|
| 2325 |
+
"step": 77200
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.08,
|
| 2329 |
+
"learning_rate": 2.3079759039496933e-05,
|
| 2330 |
+
"loss": 2.2874,
|
| 2331 |
+
"step": 77400
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 1.08,
|
| 2335 |
+
"learning_rate": 2.3010197693345762e-05,
|
| 2336 |
+
"loss": 2.2723,
|
| 2337 |
+
"step": 77600
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 1.08,
|
| 2341 |
+
"learning_rate": 2.294063634719459e-05,
|
| 2342 |
+
"loss": 2.3047,
|
| 2343 |
+
"step": 77800
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 1.09,
|
| 2347 |
+
"learning_rate": 2.287107500104342e-05,
|
| 2348 |
+
"loss": 2.2542,
|
| 2349 |
+
"step": 78000
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 1.09,
|
| 2353 |
+
"learning_rate": 2.2801513654892252e-05,
|
| 2354 |
+
"loss": 2.2426,
|
| 2355 |
+
"step": 78200
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 1.09,
|
| 2359 |
+
"learning_rate": 2.273195230874108e-05,
|
| 2360 |
+
"loss": 2.2554,
|
| 2361 |
+
"step": 78400
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 1.09,
|
| 2365 |
+
"learning_rate": 2.266239096258991e-05,
|
| 2366 |
+
"loss": 2.3179,
|
| 2367 |
+
"step": 78600
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.1,
|
| 2371 |
+
"learning_rate": 2.2592829616438738e-05,
|
| 2372 |
+
"loss": 2.2934,
|
| 2373 |
+
"step": 78800
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 1.1,
|
| 2377 |
+
"learning_rate": 2.2523268270287567e-05,
|
| 2378 |
+
"loss": 2.2707,
|
| 2379 |
+
"step": 79000
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 1.1,
|
| 2383 |
+
"learning_rate": 2.2453706924136396e-05,
|
| 2384 |
+
"loss": 2.311,
|
| 2385 |
+
"step": 79200
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 1.1,
|
| 2389 |
+
"learning_rate": 2.2384145577985228e-05,
|
| 2390 |
+
"loss": 2.3381,
|
| 2391 |
+
"step": 79400
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 1.11,
|
| 2395 |
+
"learning_rate": 2.2314584231834054e-05,
|
| 2396 |
+
"loss": 2.247,
|
| 2397 |
+
"step": 79600
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 1.11,
|
| 2401 |
+
"learning_rate": 2.2245022885682883e-05,
|
| 2402 |
+
"loss": 2.3296,
|
| 2403 |
+
"step": 79800
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 1.11,
|
| 2407 |
+
"learning_rate": 2.2175461539531715e-05,
|
| 2408 |
+
"loss": 2.2792,
|
| 2409 |
+
"step": 80000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.12,
|
| 2413 |
+
"learning_rate": 2.2105900193380544e-05,
|
| 2414 |
+
"loss": 2.3004,
|
| 2415 |
+
"step": 80200
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 1.12,
|
| 2419 |
+
"learning_rate": 2.2036338847229373e-05,
|
| 2420 |
+
"loss": 2.2845,
|
| 2421 |
+
"step": 80400
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 1.12,
|
| 2425 |
+
"learning_rate": 2.19667775010782e-05,
|
| 2426 |
+
"loss": 2.2623,
|
| 2427 |
+
"step": 80600
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 1.12,
|
| 2431 |
+
"learning_rate": 2.189721615492703e-05,
|
| 2432 |
+
"loss": 2.2849,
|
| 2433 |
+
"step": 80800
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 1.13,
|
| 2437 |
+
"learning_rate": 2.182765480877586e-05,
|
| 2438 |
+
"loss": 2.231,
|
| 2439 |
+
"step": 81000
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 1.13,
|
| 2443 |
+
"learning_rate": 2.175809346262469e-05,
|
| 2444 |
+
"loss": 2.2894,
|
| 2445 |
+
"step": 81200
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 1.13,
|
| 2449 |
+
"learning_rate": 2.168853211647352e-05,
|
| 2450 |
+
"loss": 2.2979,
|
| 2451 |
+
"step": 81400
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 1.14,
|
| 2455 |
+
"learning_rate": 2.1618970770322346e-05,
|
| 2456 |
+
"loss": 2.2961,
|
| 2457 |
+
"step": 81600
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 1.14,
|
| 2461 |
+
"learning_rate": 2.1549409424171178e-05,
|
| 2462 |
+
"loss": 2.2625,
|
| 2463 |
+
"step": 81800
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 1.14,
|
| 2467 |
+
"learning_rate": 2.1479848078020007e-05,
|
| 2468 |
+
"loss": 2.2742,
|
| 2469 |
+
"step": 82000
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 1.14,
|
| 2473 |
+
"learning_rate": 2.1410286731868836e-05,
|
| 2474 |
+
"loss": 2.2718,
|
| 2475 |
+
"step": 82200
|
| 2476 |
+
},
|
| 2477 |
+
{
|
| 2478 |
+
"epoch": 1.15,
|
| 2479 |
+
"learning_rate": 2.1340725385717668e-05,
|
| 2480 |
+
"loss": 2.2341,
|
| 2481 |
+
"step": 82400
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 1.15,
|
| 2485 |
+
"learning_rate": 2.1271164039566493e-05,
|
| 2486 |
+
"loss": 2.2424,
|
| 2487 |
+
"step": 82600
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 1.15,
|
| 2491 |
+
"learning_rate": 2.1201602693415322e-05,
|
| 2492 |
+
"loss": 2.2147,
|
| 2493 |
+
"step": 82800
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 1.15,
|
| 2497 |
+
"learning_rate": 2.1132041347264154e-05,
|
| 2498 |
+
"loss": 2.3273,
|
| 2499 |
+
"step": 83000
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 1.16,
|
| 2503 |
+
"learning_rate": 2.1062480001112983e-05,
|
| 2504 |
+
"loss": 2.2852,
|
| 2505 |
+
"step": 83200
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 1.16,
|
| 2509 |
+
"learning_rate": 2.0992918654961812e-05,
|
| 2510 |
+
"loss": 2.301,
|
| 2511 |
+
"step": 83400
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 1.16,
|
| 2515 |
+
"learning_rate": 2.092335730881064e-05,
|
| 2516 |
+
"loss": 2.2577,
|
| 2517 |
+
"step": 83600
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 1.17,
|
| 2521 |
+
"learning_rate": 2.085379596265947e-05,
|
| 2522 |
+
"loss": 2.3224,
|
| 2523 |
+
"step": 83800
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 1.17,
|
| 2527 |
+
"learning_rate": 2.07842346165083e-05,
|
| 2528 |
+
"loss": 2.2808,
|
| 2529 |
+
"step": 84000
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 1.17,
|
| 2533 |
+
"learning_rate": 2.071467327035713e-05,
|
| 2534 |
+
"loss": 2.369,
|
| 2535 |
+
"step": 84200
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 1.17,
|
| 2539 |
+
"learning_rate": 2.064511192420596e-05,
|
| 2540 |
+
"loss": 2.2545,
|
| 2541 |
+
"step": 84400
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 1.18,
|
| 2545 |
+
"learning_rate": 2.0575550578054785e-05,
|
| 2546 |
+
"loss": 2.2866,
|
| 2547 |
+
"step": 84600
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 1.18,
|
| 2551 |
+
"learning_rate": 2.0505989231903618e-05,
|
| 2552 |
+
"loss": 2.2575,
|
| 2553 |
+
"step": 84800
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 1.18,
|
| 2557 |
+
"learning_rate": 2.0436427885752446e-05,
|
| 2558 |
+
"loss": 2.2734,
|
| 2559 |
+
"step": 85000
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 1.19,
|
| 2563 |
+
"learning_rate": 2.0366866539601275e-05,
|
| 2564 |
+
"loss": 2.3246,
|
| 2565 |
+
"step": 85200
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 1.19,
|
| 2569 |
+
"learning_rate": 2.0297305193450104e-05,
|
| 2570 |
+
"loss": 2.2856,
|
| 2571 |
+
"step": 85400
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 1.19,
|
| 2575 |
+
"learning_rate": 2.0227743847298933e-05,
|
| 2576 |
+
"loss": 2.2153,
|
| 2577 |
+
"step": 85600
|
| 2578 |
+
},
|
| 2579 |
+
{
|
| 2580 |
+
"epoch": 1.19,
|
| 2581 |
+
"learning_rate": 2.0158182501147762e-05,
|
| 2582 |
+
"loss": 2.3101,
|
| 2583 |
+
"step": 85800
|
| 2584 |
+
},
|
| 2585 |
+
{
|
| 2586 |
+
"epoch": 1.2,
|
| 2587 |
+
"learning_rate": 2.0088621154996594e-05,
|
| 2588 |
+
"loss": 2.256,
|
| 2589 |
+
"step": 86000
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 1.2,
|
| 2593 |
+
"learning_rate": 2.0019059808845423e-05,
|
| 2594 |
+
"loss": 2.2453,
|
| 2595 |
+
"step": 86200
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 1.2,
|
| 2599 |
+
"learning_rate": 1.9949498462694252e-05,
|
| 2600 |
+
"loss": 2.2607,
|
| 2601 |
+
"step": 86400
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 1.2,
|
| 2605 |
+
"learning_rate": 1.987993711654308e-05,
|
| 2606 |
+
"loss": 2.2636,
|
| 2607 |
+
"step": 86600
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 1.21,
|
| 2611 |
+
"learning_rate": 1.981037577039191e-05,
|
| 2612 |
+
"loss": 2.2273,
|
| 2613 |
+
"step": 86800
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 1.21,
|
| 2617 |
+
"learning_rate": 1.974081442424074e-05,
|
| 2618 |
+
"loss": 2.2384,
|
| 2619 |
+
"step": 87000
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 1.21,
|
| 2623 |
+
"learning_rate": 1.9671253078089567e-05,
|
| 2624 |
+
"loss": 2.234,
|
| 2625 |
+
"step": 87200
|
| 2626 |
+
},
|
| 2627 |
+
{
|
| 2628 |
+
"epoch": 1.22,
|
| 2629 |
+
"learning_rate": 1.96016917319384e-05,
|
| 2630 |
+
"loss": 2.2819,
|
| 2631 |
+
"step": 87400
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 1.22,
|
| 2635 |
+
"learning_rate": 1.9532130385787225e-05,
|
| 2636 |
+
"loss": 2.2695,
|
| 2637 |
+
"step": 87600
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 1.22,
|
| 2641 |
+
"learning_rate": 1.9462569039636054e-05,
|
| 2642 |
+
"loss": 2.2836,
|
| 2643 |
+
"step": 87800
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 1.22,
|
| 2647 |
+
"learning_rate": 1.9393007693484886e-05,
|
| 2648 |
+
"loss": 2.2486,
|
| 2649 |
+
"step": 88000
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 1.23,
|
| 2653 |
+
"learning_rate": 1.9323446347333715e-05,
|
| 2654 |
+
"loss": 2.2448,
|
| 2655 |
+
"step": 88200
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 1.23,
|
| 2659 |
+
"learning_rate": 1.9253885001182544e-05,
|
| 2660 |
+
"loss": 2.3028,
|
| 2661 |
+
"step": 88400
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 1.23,
|
| 2665 |
+
"learning_rate": 1.9184323655031373e-05,
|
| 2666 |
+
"loss": 2.2855,
|
| 2667 |
+
"step": 88600
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 1.24,
|
| 2671 |
+
"learning_rate": 1.91147623088802e-05,
|
| 2672 |
+
"loss": 2.2171,
|
| 2673 |
+
"step": 88800
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 1.24,
|
| 2677 |
+
"learning_rate": 1.904520096272903e-05,
|
| 2678 |
+
"loss": 2.2834,
|
| 2679 |
+
"step": 89000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 1.24,
|
| 2683 |
+
"learning_rate": 1.8975639616577863e-05,
|
| 2684 |
+
"loss": 2.2743,
|
| 2685 |
+
"step": 89200
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 1.24,
|
| 2689 |
+
"learning_rate": 1.890607827042669e-05,
|
| 2690 |
+
"loss": 2.2437,
|
| 2691 |
+
"step": 89400
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 1.25,
|
| 2695 |
+
"learning_rate": 1.8836516924275517e-05,
|
| 2696 |
+
"loss": 2.2791,
|
| 2697 |
+
"step": 89600
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 1.25,
|
| 2701 |
+
"learning_rate": 1.876695557812435e-05,
|
| 2702 |
+
"loss": 2.2572,
|
| 2703 |
+
"step": 89800
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 1.25,
|
| 2707 |
+
"learning_rate": 1.8697394231973178e-05,
|
| 2708 |
+
"loss": 2.3147,
|
| 2709 |
+
"step": 90000
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 1.25,
|
| 2713 |
+
"learning_rate": 1.8627832885822007e-05,
|
| 2714 |
+
"loss": 2.2677,
|
| 2715 |
+
"step": 90200
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 1.26,
|
| 2719 |
+
"learning_rate": 1.855827153967084e-05,
|
| 2720 |
+
"loss": 2.2691,
|
| 2721 |
+
"step": 90400
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 1.26,
|
| 2725 |
+
"learning_rate": 1.8488710193519665e-05,
|
| 2726 |
+
"loss": 2.2629,
|
| 2727 |
+
"step": 90600
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 1.26,
|
| 2731 |
+
"learning_rate": 1.8419148847368493e-05,
|
| 2732 |
+
"loss": 2.2621,
|
| 2733 |
+
"step": 90800
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 1.27,
|
| 2737 |
+
"learning_rate": 1.8349587501217326e-05,
|
| 2738 |
+
"loss": 2.2666,
|
| 2739 |
+
"step": 91000
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 1.27,
|
| 2743 |
+
"learning_rate": 1.8280026155066155e-05,
|
| 2744 |
+
"loss": 2.2626,
|
| 2745 |
+
"step": 91200
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 1.27,
|
| 2749 |
+
"learning_rate": 1.8210464808914983e-05,
|
| 2750 |
+
"loss": 2.2453,
|
| 2751 |
+
"step": 91400
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 1.27,
|
| 2755 |
+
"learning_rate": 1.8140903462763812e-05,
|
| 2756 |
+
"loss": 2.2943,
|
| 2757 |
+
"step": 91600
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 1.28,
|
| 2761 |
+
"learning_rate": 1.807134211661264e-05,
|
| 2762 |
+
"loss": 2.2731,
|
| 2763 |
+
"step": 91800
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 1.28,
|
| 2767 |
+
"learning_rate": 1.800178077046147e-05,
|
| 2768 |
+
"loss": 2.2229,
|
| 2769 |
+
"step": 92000
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 1.28,
|
| 2773 |
+
"learning_rate": 1.7932219424310302e-05,
|
| 2774 |
+
"loss": 2.2683,
|
| 2775 |
+
"step": 92200
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 1.29,
|
| 2779 |
+
"learning_rate": 1.786265807815913e-05,
|
| 2780 |
+
"loss": 2.2642,
|
| 2781 |
+
"step": 92400
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 1.29,
|
| 2785 |
+
"learning_rate": 1.7793096732007957e-05,
|
| 2786 |
+
"loss": 2.2761,
|
| 2787 |
+
"step": 92600
|
| 2788 |
+
},
|
| 2789 |
+
{
|
| 2790 |
+
"epoch": 1.29,
|
| 2791 |
+
"learning_rate": 1.772353538585679e-05,
|
| 2792 |
+
"loss": 2.2218,
|
| 2793 |
+
"step": 92800
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 1.29,
|
| 2797 |
+
"learning_rate": 1.7653974039705618e-05,
|
| 2798 |
+
"loss": 2.2866,
|
| 2799 |
+
"step": 93000
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 1.3,
|
| 2803 |
+
"learning_rate": 1.7584412693554447e-05,
|
| 2804 |
+
"loss": 2.267,
|
| 2805 |
+
"step": 93200
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 1.3,
|
| 2809 |
+
"learning_rate": 1.751485134740328e-05,
|
| 2810 |
+
"loss": 2.2521,
|
| 2811 |
+
"step": 93400
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 1.3,
|
| 2815 |
+
"learning_rate": 1.7445290001252104e-05,
|
| 2816 |
+
"loss": 2.2212,
|
| 2817 |
+
"step": 93600
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 1.3,
|
| 2821 |
+
"learning_rate": 1.7375728655100933e-05,
|
| 2822 |
+
"loss": 2.2599,
|
| 2823 |
+
"step": 93800
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 1.31,
|
| 2827 |
+
"learning_rate": 1.7306167308949765e-05,
|
| 2828 |
+
"loss": 2.2767,
|
| 2829 |
+
"step": 94000
|
| 2830 |
+
},
|
| 2831 |
+
{
|
| 2832 |
+
"epoch": 1.31,
|
| 2833 |
+
"learning_rate": 1.7236605962798594e-05,
|
| 2834 |
+
"loss": 2.2473,
|
| 2835 |
+
"step": 94200
|
| 2836 |
+
},
|
| 2837 |
+
{
|
| 2838 |
+
"epoch": 1.31,
|
| 2839 |
+
"learning_rate": 1.716704461664742e-05,
|
| 2840 |
+
"loss": 2.2549,
|
| 2841 |
+
"step": 94400
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 1.32,
|
| 2845 |
+
"learning_rate": 1.7097483270496252e-05,
|
| 2846 |
+
"loss": 2.2694,
|
| 2847 |
+
"step": 94600
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 1.32,
|
| 2851 |
+
"learning_rate": 1.702792192434508e-05,
|
| 2852 |
+
"loss": 2.2805,
|
| 2853 |
+
"step": 94800
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 1.32,
|
| 2857 |
+
"learning_rate": 1.695836057819391e-05,
|
| 2858 |
+
"loss": 2.2728,
|
| 2859 |
+
"step": 95000
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 1.32,
|
| 2863 |
+
"learning_rate": 1.688879923204274e-05,
|
| 2864 |
+
"loss": 2.2649,
|
| 2865 |
+
"step": 95200
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 1.33,
|
| 2869 |
+
"learning_rate": 1.6819237885891567e-05,
|
| 2870 |
+
"loss": 2.2667,
|
| 2871 |
+
"step": 95400
|
| 2872 |
+
},
|
| 2873 |
+
{
|
| 2874 |
+
"epoch": 1.33,
|
| 2875 |
+
"learning_rate": 1.6749676539740396e-05,
|
| 2876 |
+
"loss": 2.2807,
|
| 2877 |
+
"step": 95600
|
| 2878 |
+
},
|
| 2879 |
+
{
|
| 2880 |
+
"epoch": 1.33,
|
| 2881 |
+
"learning_rate": 1.668011519358923e-05,
|
| 2882 |
+
"loss": 2.2694,
|
| 2883 |
+
"step": 95800
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 1.34,
|
| 2887 |
+
"learning_rate": 1.6610553847438057e-05,
|
| 2888 |
+
"loss": 2.2407,
|
| 2889 |
+
"step": 96000
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 1.34,
|
| 2893 |
+
"learning_rate": 1.6540992501286886e-05,
|
| 2894 |
+
"loss": 2.2147,
|
| 2895 |
+
"step": 96200
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 1.34,
|
| 2899 |
+
"learning_rate": 1.6471431155135715e-05,
|
| 2900 |
+
"loss": 2.266,
|
| 2901 |
+
"step": 96400
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 1.34,
|
| 2905 |
+
"learning_rate": 1.6401869808984544e-05,
|
| 2906 |
+
"loss": 2.2406,
|
| 2907 |
+
"step": 96600
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 1.35,
|
| 2911 |
+
"learning_rate": 1.6332308462833373e-05,
|
| 2912 |
+
"loss": 2.2371,
|
| 2913 |
+
"step": 96800
|
| 2914 |
+
},
|
| 2915 |
+
{
|
| 2916 |
+
"epoch": 1.35,
|
| 2917 |
+
"learning_rate": 1.62627471166822e-05,
|
| 2918 |
+
"loss": 2.2532,
|
| 2919 |
+
"step": 97000
|
| 2920 |
+
},
|
| 2921 |
+
{
|
| 2922 |
+
"epoch": 1.35,
|
| 2923 |
+
"learning_rate": 1.6193185770531034e-05,
|
| 2924 |
+
"loss": 2.2243,
|
| 2925 |
+
"step": 97200
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 1.36,
|
| 2929 |
+
"learning_rate": 1.612362442437986e-05,
|
| 2930 |
+
"loss": 2.2837,
|
| 2931 |
+
"step": 97400
|
| 2932 |
+
},
|
| 2933 |
+
{
|
| 2934 |
+
"epoch": 1.36,
|
| 2935 |
+
"learning_rate": 1.605406307822869e-05,
|
| 2936 |
+
"loss": 2.2991,
|
| 2937 |
+
"step": 97600
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 1.36,
|
| 2941 |
+
"learning_rate": 1.598450173207752e-05,
|
| 2942 |
+
"loss": 2.2166,
|
| 2943 |
+
"step": 97800
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 1.36,
|
| 2947 |
+
"learning_rate": 1.591494038592635e-05,
|
| 2948 |
+
"loss": 2.2738,
|
| 2949 |
+
"step": 98000
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 1.37,
|
| 2953 |
+
"learning_rate": 1.5845379039775178e-05,
|
| 2954 |
+
"loss": 2.2654,
|
| 2955 |
+
"step": 98200
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 1.37,
|
| 2959 |
+
"learning_rate": 1.5775817693624007e-05,
|
| 2960 |
+
"loss": 2.2906,
|
| 2961 |
+
"step": 98400
|
| 2962 |
+
},
|
| 2963 |
+
{
|
| 2964 |
+
"epoch": 1.37,
|
| 2965 |
+
"learning_rate": 1.5706256347472836e-05,
|
| 2966 |
+
"loss": 2.2701,
|
| 2967 |
+
"step": 98600
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 1.37,
|
| 2971 |
+
"learning_rate": 1.5636695001321665e-05,
|
| 2972 |
+
"loss": 2.2599,
|
| 2973 |
+
"step": 98800
|
| 2974 |
+
},
|
| 2975 |
+
{
|
| 2976 |
+
"epoch": 1.38,
|
| 2977 |
+
"learning_rate": 1.5567133655170497e-05,
|
| 2978 |
+
"loss": 2.2211,
|
| 2979 |
+
"step": 99000
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 1.38,
|
| 2983 |
+
"learning_rate": 1.5497572309019326e-05,
|
| 2984 |
+
"loss": 2.2448,
|
| 2985 |
+
"step": 99200
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 1.38,
|
| 2989 |
+
"learning_rate": 1.542801096286815e-05,
|
| 2990 |
+
"loss": 2.2691,
|
| 2991 |
+
"step": 99400
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 1.39,
|
| 2995 |
+
"learning_rate": 1.5358449616716984e-05,
|
| 2996 |
+
"loss": 2.2314,
|
| 2997 |
+
"step": 99600
|
| 2998 |
+
},
|
| 2999 |
+
{
|
| 3000 |
+
"epoch": 1.39,
|
| 3001 |
+
"learning_rate": 1.5288888270565812e-05,
|
| 3002 |
+
"loss": 2.2875,
|
| 3003 |
+
"step": 99800
|
| 3004 |
+
},
|
| 3005 |
+
{
|
| 3006 |
+
"epoch": 1.39,
|
| 3007 |
+
"learning_rate": 1.5219326924414643e-05,
|
| 3008 |
+
"loss": 2.2662,
|
| 3009 |
+
"step": 100000
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 1.39,
|
| 3013 |
+
"learning_rate": 1.5149765578263472e-05,
|
| 3014 |
+
"loss": 2.2468,
|
| 3015 |
+
"step": 100200
|
| 3016 |
+
},
|
| 3017 |
+
{
|
| 3018 |
+
"epoch": 1.4,
|
| 3019 |
+
"learning_rate": 1.5080204232112299e-05,
|
| 3020 |
+
"loss": 2.2398,
|
| 3021 |
+
"step": 100400
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 1.4,
|
| 3025 |
+
"learning_rate": 1.501064288596113e-05,
|
| 3026 |
+
"loss": 2.2485,
|
| 3027 |
+
"step": 100600
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 1.4,
|
| 3031 |
+
"learning_rate": 1.4941081539809958e-05,
|
| 3032 |
+
"loss": 2.235,
|
| 3033 |
+
"step": 100800
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 1.41,
|
| 3037 |
+
"learning_rate": 1.4871520193658789e-05,
|
| 3038 |
+
"loss": 2.2683,
|
| 3039 |
+
"step": 101000
|
| 3040 |
+
},
|
| 3041 |
+
{
|
| 3042 |
+
"epoch": 1.41,
|
| 3043 |
+
"learning_rate": 1.480195884750762e-05,
|
| 3044 |
+
"loss": 2.2748,
|
| 3045 |
+
"step": 101200
|
| 3046 |
+
},
|
| 3047 |
+
{
|
| 3048 |
+
"epoch": 1.41,
|
| 3049 |
+
"learning_rate": 1.4732397501356447e-05,
|
| 3050 |
+
"loss": 2.2722,
|
| 3051 |
+
"step": 101400
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 1.41,
|
| 3055 |
+
"learning_rate": 1.4662836155205275e-05,
|
| 3056 |
+
"loss": 2.256,
|
| 3057 |
+
"step": 101600
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 1.42,
|
| 3061 |
+
"learning_rate": 1.4593274809054106e-05,
|
| 3062 |
+
"loss": 2.2949,
|
| 3063 |
+
"step": 101800
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 1.42,
|
| 3067 |
+
"learning_rate": 1.4523713462902935e-05,
|
| 3068 |
+
"loss": 2.2466,
|
| 3069 |
+
"step": 102000
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 1.42,
|
| 3073 |
+
"learning_rate": 1.4454152116751765e-05,
|
| 3074 |
+
"loss": 2.2626,
|
| 3075 |
+
"step": 102200
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 1.42,
|
| 3079 |
+
"learning_rate": 1.4384590770600593e-05,
|
| 3080 |
+
"loss": 2.2383,
|
| 3081 |
+
"step": 102400
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 1.43,
|
| 3085 |
+
"learning_rate": 1.4315029424449421e-05,
|
| 3086 |
+
"loss": 2.2195,
|
| 3087 |
+
"step": 102600
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 1.43,
|
| 3091 |
+
"learning_rate": 1.4245468078298252e-05,
|
| 3092 |
+
"loss": 2.2826,
|
| 3093 |
+
"step": 102800
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 1.43,
|
| 3097 |
+
"learning_rate": 1.4175906732147083e-05,
|
| 3098 |
+
"loss": 2.2879,
|
| 3099 |
+
"step": 103000
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 1.44,
|
| 3103 |
+
"learning_rate": 1.4106345385995911e-05,
|
| 3104 |
+
"loss": 2.2546,
|
| 3105 |
+
"step": 103200
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 1.44,
|
| 3109 |
+
"learning_rate": 1.4036784039844739e-05,
|
| 3110 |
+
"loss": 2.3098,
|
| 3111 |
+
"step": 103400
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 1.44,
|
| 3115 |
+
"learning_rate": 1.3967222693693569e-05,
|
| 3116 |
+
"loss": 2.2196,
|
| 3117 |
+
"step": 103600
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 1.44,
|
| 3121 |
+
"learning_rate": 1.3897661347542398e-05,
|
| 3122 |
+
"loss": 2.2716,
|
| 3123 |
+
"step": 103800
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 1.45,
|
| 3127 |
+
"learning_rate": 1.3828100001391229e-05,
|
| 3128 |
+
"loss": 2.283,
|
| 3129 |
+
"step": 104000
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 1.45,
|
| 3133 |
+
"learning_rate": 1.3758538655240057e-05,
|
| 3134 |
+
"loss": 2.2332,
|
| 3135 |
+
"step": 104200
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 1.45,
|
| 3139 |
+
"learning_rate": 1.3688977309088885e-05,
|
| 3140 |
+
"loss": 2.2702,
|
| 3141 |
+
"step": 104400
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 1.46,
|
| 3145 |
+
"learning_rate": 1.3619415962937715e-05,
|
| 3146 |
+
"loss": 2.2292,
|
| 3147 |
+
"step": 104600
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 1.46,
|
| 3151 |
+
"learning_rate": 1.3549854616786544e-05,
|
| 3152 |
+
"loss": 2.2587,
|
| 3153 |
+
"step": 104800
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 1.46,
|
| 3157 |
+
"learning_rate": 1.3480293270635375e-05,
|
| 3158 |
+
"loss": 2.2908,
|
| 3159 |
+
"step": 105000
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 1.46,
|
| 3163 |
+
"learning_rate": 1.3410731924484205e-05,
|
| 3164 |
+
"loss": 2.2374,
|
| 3165 |
+
"step": 105200
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 1.47,
|
| 3169 |
+
"learning_rate": 1.3341170578333032e-05,
|
| 3170 |
+
"loss": 2.202,
|
| 3171 |
+
"step": 105400
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 1.47,
|
| 3175 |
+
"learning_rate": 1.3271609232181861e-05,
|
| 3176 |
+
"loss": 2.2496,
|
| 3177 |
+
"step": 105600
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 1.47,
|
| 3181 |
+
"learning_rate": 1.3202047886030692e-05,
|
| 3182 |
+
"loss": 2.2488,
|
| 3183 |
+
"step": 105800
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 1.47,
|
| 3187 |
+
"learning_rate": 1.313248653987952e-05,
|
| 3188 |
+
"loss": 2.2002,
|
| 3189 |
+
"step": 106000
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 1.48,
|
| 3193 |
+
"learning_rate": 1.3062925193728351e-05,
|
| 3194 |
+
"loss": 2.2189,
|
| 3195 |
+
"step": 106200
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 1.48,
|
| 3199 |
+
"learning_rate": 1.2993363847577178e-05,
|
| 3200 |
+
"loss": 2.2373,
|
| 3201 |
+
"step": 106400
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 1.48,
|
| 3205 |
+
"learning_rate": 1.2923802501426007e-05,
|
| 3206 |
+
"loss": 2.2146,
|
| 3207 |
+
"step": 106600
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 1.49,
|
| 3211 |
+
"learning_rate": 1.2854241155274838e-05,
|
| 3212 |
+
"loss": 2.2442,
|
| 3213 |
+
"step": 106800
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 1.49,
|
| 3217 |
+
"learning_rate": 1.2784679809123668e-05,
|
| 3218 |
+
"loss": 2.2113,
|
| 3219 |
+
"step": 107000
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 1.49,
|
| 3223 |
+
"learning_rate": 1.2715118462972497e-05,
|
| 3224 |
+
"loss": 2.2127,
|
| 3225 |
+
"step": 107200
|
| 3226 |
+
},
|
| 3227 |
+
{
|
| 3228 |
+
"epoch": 1.49,
|
| 3229 |
+
"learning_rate": 1.2645557116821324e-05,
|
| 3230 |
+
"loss": 2.2608,
|
| 3231 |
+
"step": 107400
|
| 3232 |
+
},
|
| 3233 |
+
{
|
| 3234 |
+
"epoch": 1.5,
|
| 3235 |
+
"learning_rate": 1.2575995770670155e-05,
|
| 3236 |
+
"loss": 2.2913,
|
| 3237 |
+
"step": 107600
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 1.5,
|
| 3241 |
+
"learning_rate": 1.2506434424518984e-05,
|
| 3242 |
+
"loss": 2.2128,
|
| 3243 |
+
"step": 107800
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 1.5,
|
| 3247 |
+
"learning_rate": 1.2436873078367812e-05,
|
| 3248 |
+
"loss": 2.2995,
|
| 3249 |
+
"step": 108000
|
| 3250 |
+
},
|
| 3251 |
+
{
|
| 3252 |
+
"epoch": 1.51,
|
| 3253 |
+
"learning_rate": 1.2367311732216643e-05,
|
| 3254 |
+
"loss": 2.1982,
|
| 3255 |
+
"step": 108200
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 1.51,
|
| 3259 |
+
"learning_rate": 1.2297750386065472e-05,
|
| 3260 |
+
"loss": 2.2296,
|
| 3261 |
+
"step": 108400
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 1.51,
|
| 3265 |
+
"learning_rate": 1.22281890399143e-05,
|
| 3266 |
+
"loss": 2.2605,
|
| 3267 |
+
"step": 108600
|
| 3268 |
+
},
|
| 3269 |
+
{
|
| 3270 |
+
"epoch": 1.51,
|
| 3271 |
+
"learning_rate": 1.2158627693763131e-05,
|
| 3272 |
+
"loss": 2.2234,
|
| 3273 |
+
"step": 108800
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 1.52,
|
| 3277 |
+
"learning_rate": 1.2089066347611958e-05,
|
| 3278 |
+
"loss": 2.2545,
|
| 3279 |
+
"step": 109000
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 1.52,
|
| 3283 |
+
"learning_rate": 1.2019505001460789e-05,
|
| 3284 |
+
"loss": 2.2522,
|
| 3285 |
+
"step": 109200
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 1.52,
|
| 3289 |
+
"learning_rate": 1.1949943655309618e-05,
|
| 3290 |
+
"loss": 2.2098,
|
| 3291 |
+
"step": 109400
|
| 3292 |
+
},
|
| 3293 |
+
{
|
| 3294 |
+
"epoch": 1.52,
|
| 3295 |
+
"learning_rate": 1.1880382309158447e-05,
|
| 3296 |
+
"loss": 2.2733,
|
| 3297 |
+
"step": 109600
|
| 3298 |
+
},
|
| 3299 |
+
{
|
| 3300 |
+
"epoch": 1.53,
|
| 3301 |
+
"learning_rate": 1.1810820963007277e-05,
|
| 3302 |
+
"loss": 2.2238,
|
| 3303 |
+
"step": 109800
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 1.53,
|
| 3307 |
+
"learning_rate": 1.1741259616856106e-05,
|
| 3308 |
+
"loss": 2.2313,
|
| 3309 |
+
"step": 110000
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 1.53,
|
| 3313 |
+
"learning_rate": 1.1671698270704935e-05,
|
| 3314 |
+
"loss": 2.2784,
|
| 3315 |
+
"step": 110200
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 1.54,
|
| 3319 |
+
"learning_rate": 1.1602136924553766e-05,
|
| 3320 |
+
"loss": 2.2708,
|
| 3321 |
+
"step": 110400
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 1.54,
|
| 3325 |
+
"learning_rate": 1.1532575578402593e-05,
|
| 3326 |
+
"loss": 2.2571,
|
| 3327 |
+
"step": 110600
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 1.54,
|
| 3331 |
+
"learning_rate": 1.1463014232251423e-05,
|
| 3332 |
+
"loss": 2.2817,
|
| 3333 |
+
"step": 110800
|
| 3334 |
+
},
|
| 3335 |
+
{
|
| 3336 |
+
"epoch": 1.54,
|
| 3337 |
+
"learning_rate": 1.1393452886100252e-05,
|
| 3338 |
+
"loss": 2.2155,
|
| 3339 |
+
"step": 111000
|
| 3340 |
+
},
|
| 3341 |
+
{
|
| 3342 |
+
"epoch": 1.55,
|
| 3343 |
+
"learning_rate": 1.1323891539949081e-05,
|
| 3344 |
+
"loss": 2.2027,
|
| 3345 |
+
"step": 111200
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 1.55,
|
| 3349 |
+
"learning_rate": 1.1254330193797912e-05,
|
| 3350 |
+
"loss": 2.2216,
|
| 3351 |
+
"step": 111400
|
| 3352 |
+
},
|
| 3353 |
+
{
|
| 3354 |
+
"epoch": 1.55,
|
| 3355 |
+
"learning_rate": 1.118476884764674e-05,
|
| 3356 |
+
"loss": 2.2222,
|
| 3357 |
+
"step": 111600
|
| 3358 |
+
},
|
| 3359 |
+
{
|
| 3360 |
+
"epoch": 1.56,
|
| 3361 |
+
"learning_rate": 1.111520750149557e-05,
|
| 3362 |
+
"loss": 2.2636,
|
| 3363 |
+
"step": 111800
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 1.56,
|
| 3367 |
+
"learning_rate": 1.1045646155344398e-05,
|
| 3368 |
+
"loss": 2.2968,
|
| 3369 |
+
"step": 112000
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 1.56,
|
| 3373 |
+
"learning_rate": 1.0976084809193229e-05,
|
| 3374 |
+
"loss": 2.3037,
|
| 3375 |
+
"step": 112200
|
| 3376 |
+
},
|
| 3377 |
+
{
|
| 3378 |
+
"epoch": 1.56,
|
| 3379 |
+
"learning_rate": 1.0906523463042057e-05,
|
| 3380 |
+
"loss": 2.2548,
|
| 3381 |
+
"step": 112400
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 1.57,
|
| 3385 |
+
"learning_rate": 1.0836962116890886e-05,
|
| 3386 |
+
"loss": 2.2708,
|
| 3387 |
+
"step": 112600
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 1.57,
|
| 3391 |
+
"learning_rate": 1.0767400770739717e-05,
|
| 3392 |
+
"loss": 2.2342,
|
| 3393 |
+
"step": 112800
|
| 3394 |
+
},
|
| 3395 |
+
{
|
| 3396 |
+
"epoch": 1.57,
|
| 3397 |
+
"learning_rate": 1.0697839424588544e-05,
|
| 3398 |
+
"loss": 2.2659,
|
| 3399 |
+
"step": 113000
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 1.57,
|
| 3403 |
+
"learning_rate": 1.0628278078437375e-05,
|
| 3404 |
+
"loss": 2.2394,
|
| 3405 |
+
"step": 113200
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 1.58,
|
| 3409 |
+
"learning_rate": 1.0558716732286205e-05,
|
| 3410 |
+
"loss": 2.29,
|
| 3411 |
+
"step": 113400
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 1.58,
|
| 3415 |
+
"learning_rate": 1.0489155386135032e-05,
|
| 3416 |
+
"loss": 2.2658,
|
| 3417 |
+
"step": 113600
|
| 3418 |
+
},
|
| 3419 |
+
{
|
| 3420 |
+
"epoch": 1.58,
|
| 3421 |
+
"learning_rate": 1.0419594039983863e-05,
|
| 3422 |
+
"loss": 2.2412,
|
| 3423 |
+
"step": 113800
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 1.59,
|
| 3427 |
+
"learning_rate": 1.0350032693832692e-05,
|
| 3428 |
+
"loss": 2.2438,
|
| 3429 |
+
"step": 114000
|
| 3430 |
+
},
|
| 3431 |
+
{
|
| 3432 |
+
"epoch": 1.59,
|
| 3433 |
+
"learning_rate": 1.028047134768152e-05,
|
| 3434 |
+
"loss": 2.2179,
|
| 3435 |
+
"step": 114200
|
| 3436 |
+
},
|
| 3437 |
+
{
|
| 3438 |
+
"epoch": 1.59,
|
| 3439 |
+
"learning_rate": 1.0210910001530351e-05,
|
| 3440 |
+
"loss": 2.2669,
|
| 3441 |
+
"step": 114400
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 1.59,
|
| 3445 |
+
"learning_rate": 1.014134865537918e-05,
|
| 3446 |
+
"loss": 2.1845,
|
| 3447 |
+
"step": 114600
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 1.6,
|
| 3451 |
+
"learning_rate": 1.0071787309228009e-05,
|
| 3452 |
+
"loss": 2.239,
|
| 3453 |
+
"step": 114800
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 1.6,
|
| 3457 |
+
"learning_rate": 1.0002225963076838e-05,
|
| 3458 |
+
"loss": 2.2727,
|
| 3459 |
+
"step": 115000
|
| 3460 |
+
},
|
| 3461 |
+
{
|
| 3462 |
+
"epoch": 1.6,
|
| 3463 |
+
"learning_rate": 9.932664616925667e-06,
|
| 3464 |
+
"loss": 2.2217,
|
| 3465 |
+
"step": 115200
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 1.61,
|
| 3469 |
+
"learning_rate": 9.863103270774495e-06,
|
| 3470 |
+
"loss": 2.2662,
|
| 3471 |
+
"step": 115400
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 1.61,
|
| 3475 |
+
"learning_rate": 9.793541924623326e-06,
|
| 3476 |
+
"loss": 2.2163,
|
| 3477 |
+
"step": 115600
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 1.61,
|
| 3481 |
+
"learning_rate": 9.723980578472155e-06,
|
| 3482 |
+
"loss": 2.242,
|
| 3483 |
+
"step": 115800
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 1.61,
|
| 3487 |
+
"learning_rate": 9.654419232320984e-06,
|
| 3488 |
+
"loss": 2.2967,
|
| 3489 |
+
"step": 116000
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 1.62,
|
| 3493 |
+
"learning_rate": 9.584857886169814e-06,
|
| 3494 |
+
"loss": 2.2005,
|
| 3495 |
+
"step": 116200
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 1.62,
|
| 3499 |
+
"learning_rate": 9.515296540018641e-06,
|
| 3500 |
+
"loss": 2.244,
|
| 3501 |
+
"step": 116400
|
| 3502 |
+
},
|
| 3503 |
+
{
|
| 3504 |
+
"epoch": 1.62,
|
| 3505 |
+
"learning_rate": 9.445735193867472e-06,
|
| 3506 |
+
"loss": 2.2641,
|
| 3507 |
+
"step": 116600
|
| 3508 |
+
},
|
| 3509 |
+
{
|
| 3510 |
+
"epoch": 1.62,
|
| 3511 |
+
"learning_rate": 9.376173847716303e-06,
|
| 3512 |
+
"loss": 2.2491,
|
| 3513 |
+
"step": 116800
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 1.63,
|
| 3517 |
+
"learning_rate": 9.30661250156513e-06,
|
| 3518 |
+
"loss": 2.1801,
|
| 3519 |
+
"step": 117000
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 1.63,
|
| 3523 |
+
"learning_rate": 9.23705115541396e-06,
|
| 3524 |
+
"loss": 2.2559,
|
| 3525 |
+
"step": 117200
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 1.63,
|
| 3529 |
+
"learning_rate": 9.167489809262789e-06,
|
| 3530 |
+
"loss": 2.2147,
|
| 3531 |
+
"step": 117400
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 1.64,
|
| 3535 |
+
"learning_rate": 9.097928463111618e-06,
|
| 3536 |
+
"loss": 2.2392,
|
| 3537 |
+
"step": 117600
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 1.64,
|
| 3541 |
+
"learning_rate": 9.028367116960449e-06,
|
| 3542 |
+
"loss": 2.2059,
|
| 3543 |
+
"step": 117800
|
| 3544 |
+
},
|
| 3545 |
+
{
|
| 3546 |
+
"epoch": 1.64,
|
| 3547 |
+
"learning_rate": 8.958805770809277e-06,
|
| 3548 |
+
"loss": 2.2391,
|
| 3549 |
+
"step": 118000
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 1.64,
|
| 3553 |
+
"learning_rate": 8.889244424658106e-06,
|
| 3554 |
+
"loss": 2.2566,
|
| 3555 |
+
"step": 118200
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 1.65,
|
| 3559 |
+
"learning_rate": 8.819683078506935e-06,
|
| 3560 |
+
"loss": 2.1766,
|
| 3561 |
+
"step": 118400
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 1.65,
|
| 3565 |
+
"learning_rate": 8.750121732355766e-06,
|
| 3566 |
+
"loss": 2.2256,
|
| 3567 |
+
"step": 118600
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 1.65,
|
| 3571 |
+
"learning_rate": 8.680560386204594e-06,
|
| 3572 |
+
"loss": 2.2554,
|
| 3573 |
+
"step": 118800
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 1.66,
|
| 3577 |
+
"learning_rate": 8.610999040053423e-06,
|
| 3578 |
+
"loss": 2.2409,
|
| 3579 |
+
"step": 119000
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 1.66,
|
| 3583 |
+
"learning_rate": 8.541437693902254e-06,
|
| 3584 |
+
"loss": 2.2004,
|
| 3585 |
+
"step": 119200
|
| 3586 |
+
},
|
| 3587 |
+
{
|
| 3588 |
+
"epoch": 1.66,
|
| 3589 |
+
"learning_rate": 8.471876347751081e-06,
|
| 3590 |
+
"loss": 2.2624,
|
| 3591 |
+
"step": 119400
|
| 3592 |
+
},
|
| 3593 |
+
{
|
| 3594 |
+
"epoch": 1.66,
|
| 3595 |
+
"learning_rate": 8.402315001599912e-06,
|
| 3596 |
+
"loss": 2.2336,
|
| 3597 |
+
"step": 119600
|
| 3598 |
+
},
|
| 3599 |
+
{
|
| 3600 |
+
"epoch": 1.67,
|
| 3601 |
+
"learning_rate": 8.33275365544874e-06,
|
| 3602 |
+
"loss": 2.2002,
|
| 3603 |
+
"step": 119800
|
| 3604 |
+
},
|
| 3605 |
+
{
|
| 3606 |
+
"epoch": 1.67,
|
| 3607 |
+
"learning_rate": 8.26319230929757e-06,
|
| 3608 |
+
"loss": 2.2472,
|
| 3609 |
+
"step": 120000
|
| 3610 |
+
},
|
| 3611 |
+
{
|
| 3612 |
+
"epoch": 1.67,
|
| 3613 |
+
"learning_rate": 8.1936309631464e-06,
|
| 3614 |
+
"loss": 2.2573,
|
| 3615 |
+
"step": 120200
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 1.68,
|
| 3619 |
+
"learning_rate": 8.124069616995229e-06,
|
| 3620 |
+
"loss": 2.2176,
|
| 3621 |
+
"step": 120400
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 1.68,
|
| 3625 |
+
"learning_rate": 8.054508270844058e-06,
|
| 3626 |
+
"loss": 2.2385,
|
| 3627 |
+
"step": 120600
|
| 3628 |
+
},
|
| 3629 |
+
{
|
| 3630 |
+
"epoch": 1.68,
|
| 3631 |
+
"learning_rate": 7.984946924692888e-06,
|
| 3632 |
+
"loss": 2.2575,
|
| 3633 |
+
"step": 120800
|
| 3634 |
+
},
|
| 3635 |
+
{
|
| 3636 |
+
"epoch": 1.68,
|
| 3637 |
+
"learning_rate": 7.915385578541715e-06,
|
| 3638 |
+
"loss": 2.2158,
|
| 3639 |
+
"step": 121000
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 1.69,
|
| 3643 |
+
"learning_rate": 7.845824232390546e-06,
|
| 3644 |
+
"loss": 2.2307,
|
| 3645 |
+
"step": 121200
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 1.69,
|
| 3649 |
+
"learning_rate": 7.776262886239375e-06,
|
| 3650 |
+
"loss": 2.2312,
|
| 3651 |
+
"step": 121400
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 1.69,
|
| 3655 |
+
"learning_rate": 7.706701540088204e-06,
|
| 3656 |
+
"loss": 2.251,
|
| 3657 |
+
"step": 121600
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 1.69,
|
| 3661 |
+
"learning_rate": 7.637140193937034e-06,
|
| 3662 |
+
"loss": 2.2864,
|
| 3663 |
+
"step": 121800
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 1.7,
|
| 3667 |
+
"learning_rate": 7.567578847785862e-06,
|
| 3668 |
+
"loss": 2.2243,
|
| 3669 |
+
"step": 122000
|
| 3670 |
+
},
|
| 3671 |
+
{
|
| 3672 |
+
"epoch": 1.7,
|
| 3673 |
+
"learning_rate": 7.498017501634693e-06,
|
| 3674 |
+
"loss": 2.252,
|
| 3675 |
+
"step": 122200
|
| 3676 |
+
},
|
| 3677 |
+
{
|
| 3678 |
+
"epoch": 1.7,
|
| 3679 |
+
"learning_rate": 7.428456155483521e-06,
|
| 3680 |
+
"loss": 2.2009,
|
| 3681 |
+
"step": 122400
|
| 3682 |
+
},
|
| 3683 |
+
{
|
| 3684 |
+
"epoch": 1.71,
|
| 3685 |
+
"learning_rate": 7.35889480933235e-06,
|
| 3686 |
+
"loss": 2.2145,
|
| 3687 |
+
"step": 122600
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 1.71,
|
| 3691 |
+
"learning_rate": 7.28933346318118e-06,
|
| 3692 |
+
"loss": 2.2335,
|
| 3693 |
+
"step": 122800
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 1.71,
|
| 3697 |
+
"learning_rate": 7.219772117030009e-06,
|
| 3698 |
+
"loss": 2.2516,
|
| 3699 |
+
"step": 123000
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 1.71,
|
| 3703 |
+
"learning_rate": 7.150210770878839e-06,
|
| 3704 |
+
"loss": 2.2146,
|
| 3705 |
+
"step": 123200
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 1.72,
|
| 3709 |
+
"learning_rate": 7.0806494247276675e-06,
|
| 3710 |
+
"loss": 2.2304,
|
| 3711 |
+
"step": 123400
|
| 3712 |
+
},
|
| 3713 |
+
{
|
| 3714 |
+
"epoch": 1.72,
|
| 3715 |
+
"learning_rate": 7.011088078576497e-06,
|
| 3716 |
+
"loss": 2.1978,
|
| 3717 |
+
"step": 123600
|
| 3718 |
+
},
|
| 3719 |
+
{
|
| 3720 |
+
"epoch": 1.72,
|
| 3721 |
+
"learning_rate": 6.941526732425327e-06,
|
| 3722 |
+
"loss": 2.2488,
|
| 3723 |
+
"step": 123800
|
| 3724 |
+
},
|
| 3725 |
+
{
|
| 3726 |
+
"epoch": 1.73,
|
| 3727 |
+
"learning_rate": 6.871965386274155e-06,
|
| 3728 |
+
"loss": 2.2347,
|
| 3729 |
+
"step": 124000
|
| 3730 |
+
},
|
| 3731 |
+
{
|
| 3732 |
+
"epoch": 1.73,
|
| 3733 |
+
"learning_rate": 6.8024040401229855e-06,
|
| 3734 |
+
"loss": 2.2496,
|
| 3735 |
+
"step": 124200
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 1.73,
|
| 3739 |
+
"learning_rate": 6.7328426939718135e-06,
|
| 3740 |
+
"loss": 2.2366,
|
| 3741 |
+
"step": 124400
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 1.73,
|
| 3745 |
+
"learning_rate": 6.663281347820643e-06,
|
| 3746 |
+
"loss": 2.1691,
|
| 3747 |
+
"step": 124600
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 1.74,
|
| 3751 |
+
"learning_rate": 6.593720001669472e-06,
|
| 3752 |
+
"loss": 2.205,
|
| 3753 |
+
"step": 124800
|
| 3754 |
+
},
|
| 3755 |
+
{
|
| 3756 |
+
"epoch": 1.74,
|
| 3757 |
+
"learning_rate": 6.524158655518302e-06,
|
| 3758 |
+
"loss": 2.2107,
|
| 3759 |
+
"step": 125000
|
| 3760 |
+
},
|
| 3761 |
+
{
|
| 3762 |
+
"epoch": 1.74,
|
| 3763 |
+
"learning_rate": 6.4545973093671315e-06,
|
| 3764 |
+
"loss": 2.2188,
|
| 3765 |
+
"step": 125200
|
| 3766 |
+
},
|
| 3767 |
+
{
|
| 3768 |
+
"epoch": 1.74,
|
| 3769 |
+
"learning_rate": 6.38503596321596e-06,
|
| 3770 |
+
"loss": 2.2381,
|
| 3771 |
+
"step": 125400
|
| 3772 |
+
},
|
| 3773 |
+
{
|
| 3774 |
+
"epoch": 1.75,
|
| 3775 |
+
"learning_rate": 6.31547461706479e-06,
|
| 3776 |
+
"loss": 2.2638,
|
| 3777 |
+
"step": 125600
|
| 3778 |
+
},
|
| 3779 |
+
{
|
| 3780 |
+
"epoch": 1.75,
|
| 3781 |
+
"learning_rate": 6.245913270913619e-06,
|
| 3782 |
+
"loss": 2.2688,
|
| 3783 |
+
"step": 125800
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 1.75,
|
| 3787 |
+
"learning_rate": 6.176351924762449e-06,
|
| 3788 |
+
"loss": 2.1799,
|
| 3789 |
+
"step": 126000
|
| 3790 |
+
},
|
| 3791 |
+
{
|
| 3792 |
+
"epoch": 1.76,
|
| 3793 |
+
"learning_rate": 6.1067905786112775e-06,
|
| 3794 |
+
"loss": 2.2184,
|
| 3795 |
+
"step": 126200
|
| 3796 |
+
},
|
| 3797 |
+
{
|
| 3798 |
+
"epoch": 1.76,
|
| 3799 |
+
"learning_rate": 6.037229232460106e-06,
|
| 3800 |
+
"loss": 2.2297,
|
| 3801 |
+
"step": 126400
|
| 3802 |
+
},
|
| 3803 |
+
{
|
| 3804 |
+
"epoch": 1.76,
|
| 3805 |
+
"learning_rate": 5.967667886308936e-06,
|
| 3806 |
+
"loss": 2.2882,
|
| 3807 |
+
"step": 126600
|
| 3808 |
+
},
|
| 3809 |
+
{
|
| 3810 |
+
"epoch": 1.76,
|
| 3811 |
+
"learning_rate": 5.898106540157766e-06,
|
| 3812 |
+
"loss": 2.2388,
|
| 3813 |
+
"step": 126800
|
| 3814 |
+
},
|
| 3815 |
+
{
|
| 3816 |
+
"epoch": 1.77,
|
| 3817 |
+
"learning_rate": 5.828545194006595e-06,
|
| 3818 |
+
"loss": 2.2356,
|
| 3819 |
+
"step": 127000
|
| 3820 |
+
},
|
| 3821 |
+
{
|
| 3822 |
+
"epoch": 1.77,
|
| 3823 |
+
"learning_rate": 5.7589838478554234e-06,
|
| 3824 |
+
"loss": 2.2427,
|
| 3825 |
+
"step": 127200
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 1.77,
|
| 3829 |
+
"learning_rate": 5.689422501704253e-06,
|
| 3830 |
+
"loss": 2.2371,
|
| 3831 |
+
"step": 127400
|
| 3832 |
+
},
|
| 3833 |
+
{
|
| 3834 |
+
"epoch": 1.78,
|
| 3835 |
+
"learning_rate": 5.619861155553083e-06,
|
| 3836 |
+
"loss": 2.2639,
|
| 3837 |
+
"step": 127600
|
| 3838 |
+
},
|
| 3839 |
+
{
|
| 3840 |
+
"epoch": 1.78,
|
| 3841 |
+
"learning_rate": 5.550299809401912e-06,
|
| 3842 |
+
"loss": 2.2532,
|
| 3843 |
+
"step": 127800
|
| 3844 |
+
},
|
| 3845 |
+
{
|
| 3846 |
+
"epoch": 1.78,
|
| 3847 |
+
"learning_rate": 5.480738463250741e-06,
|
| 3848 |
+
"loss": 2.2085,
|
| 3849 |
+
"step": 128000
|
| 3850 |
+
},
|
| 3851 |
+
{
|
| 3852 |
+
"epoch": 1.78,
|
| 3853 |
+
"learning_rate": 5.41117711709957e-06,
|
| 3854 |
+
"loss": 2.263,
|
| 3855 |
+
"step": 128200
|
| 3856 |
+
},
|
| 3857 |
+
{
|
| 3858 |
+
"epoch": 1.79,
|
| 3859 |
+
"learning_rate": 5.341615770948399e-06,
|
| 3860 |
+
"loss": 2.1695,
|
| 3861 |
+
"step": 128400
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 1.79,
|
| 3865 |
+
"learning_rate": 5.272054424797229e-06,
|
| 3866 |
+
"loss": 2.2055,
|
| 3867 |
+
"step": 128600
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 1.79,
|
| 3871 |
+
"learning_rate": 5.2024930786460585e-06,
|
| 3872 |
+
"loss": 2.2349,
|
| 3873 |
+
"step": 128800
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 1.79,
|
| 3877 |
+
"learning_rate": 5.132931732494887e-06,
|
| 3878 |
+
"loss": 2.2024,
|
| 3879 |
+
"step": 129000
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 1.8,
|
| 3883 |
+
"learning_rate": 5.063370386343716e-06,
|
| 3884 |
+
"loss": 2.2263,
|
| 3885 |
+
"step": 129200
|
| 3886 |
+
},
|
| 3887 |
+
{
|
| 3888 |
+
"epoch": 1.8,
|
| 3889 |
+
"learning_rate": 4.993809040192546e-06,
|
| 3890 |
+
"loss": 2.2084,
|
| 3891 |
+
"step": 129400
|
| 3892 |
+
},
|
| 3893 |
+
{
|
| 3894 |
+
"epoch": 1.8,
|
| 3895 |
+
"learning_rate": 4.924247694041376e-06,
|
| 3896 |
+
"loss": 2.2397,
|
| 3897 |
+
"step": 129600
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 1.81,
|
| 3901 |
+
"learning_rate": 4.8546863478902045e-06,
|
| 3902 |
+
"loss": 2.2455,
|
| 3903 |
+
"step": 129800
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 1.81,
|
| 3907 |
+
"learning_rate": 4.785125001739034e-06,
|
| 3908 |
+
"loss": 2.2159,
|
| 3909 |
+
"step": 130000
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 1.81,
|
| 3913 |
+
"learning_rate": 4.715563655587863e-06,
|
| 3914 |
+
"loss": 2.1978,
|
| 3915 |
+
"step": 130200
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 1.81,
|
| 3919 |
+
"learning_rate": 4.646002309436692e-06,
|
| 3920 |
+
"loss": 2.2635,
|
| 3921 |
+
"step": 130400
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 1.82,
|
| 3925 |
+
"learning_rate": 4.576440963285522e-06,
|
| 3926 |
+
"loss": 2.2005,
|
| 3927 |
+
"step": 130600
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 1.82,
|
| 3931 |
+
"learning_rate": 4.506879617134351e-06,
|
| 3932 |
+
"loss": 2.2388,
|
| 3933 |
+
"step": 130800
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 1.82,
|
| 3937 |
+
"learning_rate": 4.43731827098318e-06,
|
| 3938 |
+
"loss": 2.2142,
|
| 3939 |
+
"step": 131000
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 1.83,
|
| 3943 |
+
"learning_rate": 4.36775692483201e-06,
|
| 3944 |
+
"loss": 2.2437,
|
| 3945 |
+
"step": 131200
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 1.83,
|
| 3949 |
+
"learning_rate": 4.298195578680839e-06,
|
| 3950 |
+
"loss": 2.2269,
|
| 3951 |
+
"step": 131400
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 1.83,
|
| 3955 |
+
"learning_rate": 4.228634232529668e-06,
|
| 3956 |
+
"loss": 2.1651,
|
| 3957 |
+
"step": 131600
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 1.83,
|
| 3961 |
+
"learning_rate": 4.159072886378497e-06,
|
| 3962 |
+
"loss": 2.2552,
|
| 3963 |
+
"step": 131800
|
| 3964 |
+
},
|
| 3965 |
+
{
|
| 3966 |
+
"epoch": 1.84,
|
| 3967 |
+
"learning_rate": 4.089511540227327e-06,
|
| 3968 |
+
"loss": 2.2328,
|
| 3969 |
+
"step": 132000
|
| 3970 |
+
},
|
| 3971 |
+
{
|
| 3972 |
+
"epoch": 1.84,
|
| 3973 |
+
"learning_rate": 4.019950194076156e-06,
|
| 3974 |
+
"loss": 2.2267,
|
| 3975 |
+
"step": 132200
|
| 3976 |
+
},
|
| 3977 |
+
{
|
| 3978 |
+
"epoch": 1.84,
|
| 3979 |
+
"learning_rate": 3.950388847924985e-06,
|
| 3980 |
+
"loss": 2.138,
|
| 3981 |
+
"step": 132400
|
| 3982 |
+
},
|
| 3983 |
+
{
|
| 3984 |
+
"epoch": 1.84,
|
| 3985 |
+
"learning_rate": 3.8808275017738145e-06,
|
| 3986 |
+
"loss": 2.1773,
|
| 3987 |
+
"step": 132600
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 1.85,
|
| 3991 |
+
"learning_rate": 3.811266155622644e-06,
|
| 3992 |
+
"loss": 2.1719,
|
| 3993 |
+
"step": 132800
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 1.85,
|
| 3997 |
+
"learning_rate": 3.741704809471473e-06,
|
| 3998 |
+
"loss": 2.2124,
|
| 3999 |
+
"step": 133000
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 1.85,
|
| 4003 |
+
"learning_rate": 3.6721434633203023e-06,
|
| 4004 |
+
"loss": 2.2293,
|
| 4005 |
+
"step": 133200
|
| 4006 |
+
},
|
| 4007 |
+
{
|
| 4008 |
+
"epoch": 1.86,
|
| 4009 |
+
"learning_rate": 3.6025821171691316e-06,
|
| 4010 |
+
"loss": 2.1802,
|
| 4011 |
+
"step": 133400
|
| 4012 |
+
},
|
| 4013 |
+
{
|
| 4014 |
+
"epoch": 1.86,
|
| 4015 |
+
"learning_rate": 3.533020771017961e-06,
|
| 4016 |
+
"loss": 2.1823,
|
| 4017 |
+
"step": 133600
|
| 4018 |
+
},
|
| 4019 |
+
{
|
| 4020 |
+
"epoch": 1.86,
|
| 4021 |
+
"learning_rate": 3.4634594248667906e-06,
|
| 4022 |
+
"loss": 2.2114,
|
| 4023 |
+
"step": 133800
|
| 4024 |
+
},
|
| 4025 |
+
{
|
| 4026 |
+
"epoch": 1.86,
|
| 4027 |
+
"learning_rate": 3.39389807871562e-06,
|
| 4028 |
+
"loss": 2.2102,
|
| 4029 |
+
"step": 134000
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 1.87,
|
| 4033 |
+
"learning_rate": 3.3243367325644487e-06,
|
| 4034 |
+
"loss": 2.2313,
|
| 4035 |
+
"step": 134200
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 1.87,
|
| 4039 |
+
"learning_rate": 3.254775386413278e-06,
|
| 4040 |
+
"loss": 2.2214,
|
| 4041 |
+
"step": 134400
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 1.87,
|
| 4045 |
+
"learning_rate": 3.1852140402621073e-06,
|
| 4046 |
+
"loss": 2.2491,
|
| 4047 |
+
"step": 134600
|
| 4048 |
+
},
|
| 4049 |
+
{
|
| 4050 |
+
"epoch": 1.88,
|
| 4051 |
+
"learning_rate": 3.1156526941109365e-06,
|
| 4052 |
+
"loss": 2.1957,
|
| 4053 |
+
"step": 134800
|
| 4054 |
+
},
|
| 4055 |
+
{
|
| 4056 |
+
"epoch": 1.88,
|
| 4057 |
+
"learning_rate": 3.046091347959766e-06,
|
| 4058 |
+
"loss": 2.251,
|
| 4059 |
+
"step": 135000
|
| 4060 |
+
},
|
| 4061 |
+
{
|
| 4062 |
+
"epoch": 1.88,
|
| 4063 |
+
"learning_rate": 2.976530001808595e-06,
|
| 4064 |
+
"loss": 2.1941,
|
| 4065 |
+
"step": 135200
|
| 4066 |
+
},
|
| 4067 |
+
{
|
| 4068 |
+
"epoch": 1.88,
|
| 4069 |
+
"learning_rate": 2.9069686556574244e-06,
|
| 4070 |
+
"loss": 2.231,
|
| 4071 |
+
"step": 135400
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 1.89,
|
| 4075 |
+
"learning_rate": 2.8374073095062537e-06,
|
| 4076 |
+
"loss": 2.1893,
|
| 4077 |
+
"step": 135600
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 1.89,
|
| 4081 |
+
"learning_rate": 2.767845963355083e-06,
|
| 4082 |
+
"loss": 2.2114,
|
| 4083 |
+
"step": 135800
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 1.89,
|
| 4087 |
+
"learning_rate": 2.6982846172039122e-06,
|
| 4088 |
+
"loss": 2.2375,
|
| 4089 |
+
"step": 136000
|
| 4090 |
+
},
|
| 4091 |
+
{
|
| 4092 |
+
"epoch": 1.89,
|
| 4093 |
+
"learning_rate": 2.6287232710527415e-06,
|
| 4094 |
+
"loss": 2.1917,
|
| 4095 |
+
"step": 136200
|
| 4096 |
+
},
|
| 4097 |
+
{
|
| 4098 |
+
"epoch": 1.9,
|
| 4099 |
+
"learning_rate": 2.559161924901571e-06,
|
| 4100 |
+
"loss": 2.2227,
|
| 4101 |
+
"step": 136400
|
| 4102 |
+
},
|
| 4103 |
+
{
|
| 4104 |
+
"epoch": 1.9,
|
| 4105 |
+
"learning_rate": 2.4896005787504e-06,
|
| 4106 |
+
"loss": 2.2534,
|
| 4107 |
+
"step": 136600
|
| 4108 |
+
},
|
| 4109 |
+
{
|
| 4110 |
+
"epoch": 1.9,
|
| 4111 |
+
"learning_rate": 2.4200392325992294e-06,
|
| 4112 |
+
"loss": 2.2529,
|
| 4113 |
+
"step": 136800
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 1.91,
|
| 4117 |
+
"learning_rate": 2.3504778864480586e-06,
|
| 4118 |
+
"loss": 2.2046,
|
| 4119 |
+
"step": 137000
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 1.91,
|
| 4123 |
+
"learning_rate": 2.280916540296888e-06,
|
| 4124 |
+
"loss": 2.262,
|
| 4125 |
+
"step": 137200
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 1.91,
|
| 4129 |
+
"learning_rate": 2.211355194145717e-06,
|
| 4130 |
+
"loss": 2.1998,
|
| 4131 |
+
"step": 137400
|
| 4132 |
+
},
|
| 4133 |
+
{
|
| 4134 |
+
"epoch": 1.91,
|
| 4135 |
+
"learning_rate": 2.1417938479945465e-06,
|
| 4136 |
+
"loss": 2.2093,
|
| 4137 |
+
"step": 137600
|
| 4138 |
+
},
|
| 4139 |
+
{
|
| 4140 |
+
"epoch": 1.92,
|
| 4141 |
+
"learning_rate": 2.0722325018433758e-06,
|
| 4142 |
+
"loss": 2.204,
|
| 4143 |
+
"step": 137800
|
| 4144 |
+
},
|
| 4145 |
+
{
|
| 4146 |
+
"epoch": 1.92,
|
| 4147 |
+
"learning_rate": 2.002671155692205e-06,
|
| 4148 |
+
"loss": 2.234,
|
| 4149 |
+
"step": 138000
|
| 4150 |
+
},
|
| 4151 |
+
{
|
| 4152 |
+
"epoch": 1.92,
|
| 4153 |
+
"learning_rate": 1.9331098095410343e-06,
|
| 4154 |
+
"loss": 2.2064,
|
| 4155 |
+
"step": 138200
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 1.93,
|
| 4159 |
+
"learning_rate": 1.8635484633898636e-06,
|
| 4160 |
+
"loss": 2.2288,
|
| 4161 |
+
"step": 138400
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 1.93,
|
| 4165 |
+
"learning_rate": 1.7939871172386929e-06,
|
| 4166 |
+
"loss": 2.1994,
|
| 4167 |
+
"step": 138600
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 1.93,
|
| 4171 |
+
"learning_rate": 1.7244257710875224e-06,
|
| 4172 |
+
"loss": 2.2235,
|
| 4173 |
+
"step": 138800
|
| 4174 |
+
},
|
| 4175 |
+
{
|
| 4176 |
+
"epoch": 1.93,
|
| 4177 |
+
"learning_rate": 1.6548644249363515e-06,
|
| 4178 |
+
"loss": 2.2251,
|
| 4179 |
+
"step": 139000
|
| 4180 |
+
},
|
| 4181 |
+
{
|
| 4182 |
+
"epoch": 1.94,
|
| 4183 |
+
"learning_rate": 1.5853030787851805e-06,
|
| 4184 |
+
"loss": 2.2068,
|
| 4185 |
+
"step": 139200
|
| 4186 |
+
},
|
| 4187 |
+
{
|
| 4188 |
+
"epoch": 1.94,
|
| 4189 |
+
"learning_rate": 1.51574173263401e-06,
|
| 4190 |
+
"loss": 2.1716,
|
| 4191 |
+
"step": 139400
|
| 4192 |
+
},
|
| 4193 |
+
{
|
| 4194 |
+
"epoch": 1.94,
|
| 4195 |
+
"learning_rate": 1.4461803864828393e-06,
|
| 4196 |
+
"loss": 2.229,
|
| 4197 |
+
"step": 139600
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 1.94,
|
| 4201 |
+
"learning_rate": 1.3766190403316686e-06,
|
| 4202 |
+
"loss": 2.2169,
|
| 4203 |
+
"step": 139800
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 1.95,
|
| 4207 |
+
"learning_rate": 1.3070576941804979e-06,
|
| 4208 |
+
"loss": 2.1852,
|
| 4209 |
+
"step": 140000
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 1.95,
|
| 4213 |
+
"learning_rate": 1.2374963480293271e-06,
|
| 4214 |
+
"loss": 2.2213,
|
| 4215 |
+
"step": 140200
|
| 4216 |
+
},
|
| 4217 |
+
{
|
| 4218 |
+
"epoch": 1.95,
|
| 4219 |
+
"learning_rate": 1.1679350018781564e-06,
|
| 4220 |
+
"loss": 2.2633,
|
| 4221 |
+
"step": 140400
|
| 4222 |
+
},
|
| 4223 |
+
{
|
| 4224 |
+
"epoch": 1.96,
|
| 4225 |
+
"learning_rate": 1.0983736557269857e-06,
|
| 4226 |
+
"loss": 2.1825,
|
| 4227 |
+
"step": 140600
|
| 4228 |
+
},
|
| 4229 |
+
{
|
| 4230 |
+
"epoch": 1.96,
|
| 4231 |
+
"learning_rate": 1.028812309575815e-06,
|
| 4232 |
+
"loss": 2.2269,
|
| 4233 |
+
"step": 140800
|
| 4234 |
+
},
|
| 4235 |
+
{
|
| 4236 |
+
"epoch": 1.96,
|
| 4237 |
+
"learning_rate": 9.592509634246443e-07,
|
| 4238 |
+
"loss": 2.2141,
|
| 4239 |
+
"step": 141000
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 1.96,
|
| 4243 |
+
"learning_rate": 8.896896172734734e-07,
|
| 4244 |
+
"loss": 2.2351,
|
| 4245 |
+
"step": 141200
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 1.97,
|
| 4249 |
+
"learning_rate": 8.201282711223028e-07,
|
| 4250 |
+
"loss": 2.2102,
|
| 4251 |
+
"step": 141400
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 1.97,
|
| 4255 |
+
"learning_rate": 7.505669249711321e-07,
|
| 4256 |
+
"loss": 2.1856,
|
| 4257 |
+
"step": 141600
|
| 4258 |
+
},
|
| 4259 |
+
{
|
| 4260 |
+
"epoch": 1.97,
|
| 4261 |
+
"learning_rate": 6.810055788199613e-07,
|
| 4262 |
+
"loss": 2.2066,
|
| 4263 |
+
"step": 141800
|
| 4264 |
+
},
|
| 4265 |
+
{
|
| 4266 |
+
"epoch": 1.98,
|
| 4267 |
+
"learning_rate": 6.114442326687907e-07,
|
| 4268 |
+
"loss": 2.2295,
|
| 4269 |
+
"step": 142000
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 1.98,
|
| 4273 |
+
"learning_rate": 5.418828865176198e-07,
|
| 4274 |
+
"loss": 2.1887,
|
| 4275 |
+
"step": 142200
|
| 4276 |
+
},
|
| 4277 |
+
{
|
| 4278 |
+
"epoch": 1.98,
|
| 4279 |
+
"learning_rate": 4.7232154036644923e-07,
|
| 4280 |
+
"loss": 2.1985,
|
| 4281 |
+
"step": 142400
|
| 4282 |
+
},
|
| 4283 |
+
{
|
| 4284 |
+
"epoch": 1.98,
|
| 4285 |
+
"learning_rate": 4.0276019421527846e-07,
|
| 4286 |
+
"loss": 2.2791,
|
| 4287 |
+
"step": 142600
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 1.99,
|
| 4291 |
+
"learning_rate": 3.3319884806410774e-07,
|
| 4292 |
+
"loss": 2.2265,
|
| 4293 |
+
"step": 142800
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 1.99,
|
| 4297 |
+
"learning_rate": 2.63637501912937e-07,
|
| 4298 |
+
"loss": 2.2137,
|
| 4299 |
+
"step": 143000
|
| 4300 |
+
},
|
| 4301 |
+
{
|
| 4302 |
+
"epoch": 1.99,
|
| 4303 |
+
"learning_rate": 1.940761557617663e-07,
|
| 4304 |
+
"loss": 2.253,
|
| 4305 |
+
"step": 143200
|
| 4306 |
+
},
|
| 4307 |
+
{
|
| 4308 |
+
"epoch": 2.0,
|
| 4309 |
+
"learning_rate": 1.2451480961059558e-07,
|
| 4310 |
+
"loss": 2.2167,
|
| 4311 |
+
"step": 143400
|
| 4312 |
+
},
|
| 4313 |
+
{
|
| 4314 |
+
"epoch": 2.0,
|
| 4315 |
+
"learning_rate": 5.495346345942487e-08,
|
| 4316 |
+
"loss": 2.2104,
|
| 4317 |
+
"step": 143600
|
| 4318 |
+
},
|
| 4319 |
{
|
| 4320 |
"epoch": 2.0,
|
| 4321 |
+
"step": 143758,
|
| 4322 |
+
"total_flos": 1.502524253059154e+17,
|
| 4323 |
+
"train_loss": 2.352252098808129,
|
| 4324 |
+
"train_runtime": 164976.46,
|
| 4325 |
+
"train_samples_per_second": 3.486,
|
| 4326 |
+
"train_steps_per_second": 0.871
|
| 4327 |
}
|
| 4328 |
],
|
| 4329 |
"logging_steps": 200,
|
| 4330 |
+
"max_steps": 143758,
|
| 4331 |
"num_input_tokens_seen": 0,
|
| 4332 |
"num_train_epochs": 2,
|
| 4333 |
"save_steps": 500,
|
| 4334 |
+
"total_flos": 1.502524253059154e+17,
|
| 4335 |
"train_batch_size": 4,
|
| 4336 |
"trial_name": null,
|
| 4337 |
"trial_params": null
|