End of training
Browse files- README.md +2 -1
- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +3332 -0
- training_loss.png +0 -0
README.md
CHANGED
|
@@ -4,6 +4,7 @@ license: apache-2.0
|
|
| 4 |
base_model: Qwen/Qwen2.5-7B-Instruct
|
| 5 |
tags:
|
| 6 |
- llama-factory
|
|
|
|
| 7 |
- generated_from_trainer
|
| 8 |
model-index:
|
| 9 |
- name: no_pipeline_science_300k
|
|
@@ -15,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# no_pipeline_science_300k
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) on
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
|
|
|
| 4 |
base_model: Qwen/Qwen2.5-7B-Instruct
|
| 5 |
tags:
|
| 6 |
- llama-factory
|
| 7 |
+
- full
|
| 8 |
- generated_from_trainer
|
| 9 |
model-index:
|
| 10 |
- name: no_pipeline_science_300k
|
|
|
|
| 16 |
|
| 17 |
# no_pipeline_science_300k
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) on the mlfoundations-dev/no_pipeline_science_300k dataset.
|
| 20 |
|
| 21 |
## Model description
|
| 22 |
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.966974900924702,
|
| 3 |
+
"total_flos": 1.2094498954511647e+19,
|
| 4 |
+
"train_loss": 0.5709579251548077,
|
| 5 |
+
"train_runtime": 47190.5732,
|
| 6 |
+
"train_samples_per_second": 5.131,
|
| 7 |
+
"train_steps_per_second": 0.01
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.966974900924702,
|
| 3 |
+
"total_flos": 1.2094498954511647e+19,
|
| 4 |
+
"train_loss": 0.5709579251548077,
|
| 5 |
+
"train_runtime": 47190.5732,
|
| 6 |
+
"train_samples_per_second": 5.131,
|
| 7 |
+
"train_steps_per_second": 0.01
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3332 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 4.966974900924702,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 470,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.010568031704095112,
|
| 13 |
+
"grad_norm": 7.083772151723263,
|
| 14 |
+
"learning_rate": 1.7021276595744682e-06,
|
| 15 |
+
"loss": 1.1953,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.021136063408190225,
|
| 20 |
+
"grad_norm": 7.131321562784017,
|
| 21 |
+
"learning_rate": 3.4042553191489363e-06,
|
| 22 |
+
"loss": 1.1964,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.031704095112285335,
|
| 27 |
+
"grad_norm": 7.0823963229107525,
|
| 28 |
+
"learning_rate": 5.106382978723404e-06,
|
| 29 |
+
"loss": 1.2092,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.04227212681638045,
|
| 34 |
+
"grad_norm": 5.141917283385601,
|
| 35 |
+
"learning_rate": 6.808510638297873e-06,
|
| 36 |
+
"loss": 1.1295,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.052840158520475564,
|
| 41 |
+
"grad_norm": 2.56300415286007,
|
| 42 |
+
"learning_rate": 8.510638297872341e-06,
|
| 43 |
+
"loss": 1.077,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.06340819022457067,
|
| 48 |
+
"grad_norm": 4.615211256351488,
|
| 49 |
+
"learning_rate": 1.0212765957446808e-05,
|
| 50 |
+
"loss": 1.0452,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.07397622192866579,
|
| 55 |
+
"grad_norm": 4.783764272721008,
|
| 56 |
+
"learning_rate": 1.1914893617021277e-05,
|
| 57 |
+
"loss": 1.0356,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0845442536327609,
|
| 62 |
+
"grad_norm": 3.8154182115223816,
|
| 63 |
+
"learning_rate": 1.3617021276595745e-05,
|
| 64 |
+
"loss": 0.9744,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.095112285336856,
|
| 69 |
+
"grad_norm": 3.8637000758137723,
|
| 70 |
+
"learning_rate": 1.5319148936170214e-05,
|
| 71 |
+
"loss": 0.9714,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.10568031704095113,
|
| 76 |
+
"grad_norm": 2.366476473063925,
|
| 77 |
+
"learning_rate": 1.7021276595744682e-05,
|
| 78 |
+
"loss": 0.9574,
|
| 79 |
+
"step": 10
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.11624834874504623,
|
| 83 |
+
"grad_norm": 2.994957368837667,
|
| 84 |
+
"learning_rate": 1.872340425531915e-05,
|
| 85 |
+
"loss": 0.9216,
|
| 86 |
+
"step": 11
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.12681638044914134,
|
| 90 |
+
"grad_norm": 2.6457297468520014,
|
| 91 |
+
"learning_rate": 2.0425531914893616e-05,
|
| 92 |
+
"loss": 0.9039,
|
| 93 |
+
"step": 12
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.13738441215323646,
|
| 97 |
+
"grad_norm": 1.6507084509279148,
|
| 98 |
+
"learning_rate": 2.2127659574468088e-05,
|
| 99 |
+
"loss": 0.8965,
|
| 100 |
+
"step": 13
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.14795244385733158,
|
| 104 |
+
"grad_norm": 1.726225549814637,
|
| 105 |
+
"learning_rate": 2.3829787234042553e-05,
|
| 106 |
+
"loss": 0.8849,
|
| 107 |
+
"step": 14
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.15852047556142668,
|
| 111 |
+
"grad_norm": 1.6588209807553262,
|
| 112 |
+
"learning_rate": 2.5531914893617025e-05,
|
| 113 |
+
"loss": 0.871,
|
| 114 |
+
"step": 15
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.1690885072655218,
|
| 118 |
+
"grad_norm": 1.2143860211983681,
|
| 119 |
+
"learning_rate": 2.723404255319149e-05,
|
| 120 |
+
"loss": 0.8617,
|
| 121 |
+
"step": 16
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.17965653896961692,
|
| 125 |
+
"grad_norm": 1.2868680178333174,
|
| 126 |
+
"learning_rate": 2.8936170212765963e-05,
|
| 127 |
+
"loss": 0.8567,
|
| 128 |
+
"step": 17
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.190224570673712,
|
| 132 |
+
"grad_norm": 1.2037257451837027,
|
| 133 |
+
"learning_rate": 3.063829787234043e-05,
|
| 134 |
+
"loss": 0.8434,
|
| 135 |
+
"step": 18
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.20079260237780713,
|
| 139 |
+
"grad_norm": 1.041214478985964,
|
| 140 |
+
"learning_rate": 3.234042553191489e-05,
|
| 141 |
+
"loss": 0.8371,
|
| 142 |
+
"step": 19
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.21136063408190225,
|
| 146 |
+
"grad_norm": 0.9037174605431649,
|
| 147 |
+
"learning_rate": 3.4042553191489365e-05,
|
| 148 |
+
"loss": 0.825,
|
| 149 |
+
"step": 20
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.22192866578599735,
|
| 153 |
+
"grad_norm": 1.3140952321772827,
|
| 154 |
+
"learning_rate": 3.574468085106383e-05,
|
| 155 |
+
"loss": 0.8314,
|
| 156 |
+
"step": 21
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.23249669749009247,
|
| 160 |
+
"grad_norm": 1.3446525845289001,
|
| 161 |
+
"learning_rate": 3.74468085106383e-05,
|
| 162 |
+
"loss": 0.8274,
|
| 163 |
+
"step": 22
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.2430647291941876,
|
| 167 |
+
"grad_norm": 1.202878622232685,
|
| 168 |
+
"learning_rate": 3.914893617021277e-05,
|
| 169 |
+
"loss": 0.8126,
|
| 170 |
+
"step": 23
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.2536327608982827,
|
| 174 |
+
"grad_norm": 0.9767090732567484,
|
| 175 |
+
"learning_rate": 4.085106382978723e-05,
|
| 176 |
+
"loss": 0.806,
|
| 177 |
+
"step": 24
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.26420079260237783,
|
| 181 |
+
"grad_norm": 1.7139214140351846,
|
| 182 |
+
"learning_rate": 4.2553191489361704e-05,
|
| 183 |
+
"loss": 0.8059,
|
| 184 |
+
"step": 25
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.2747688243064729,
|
| 188 |
+
"grad_norm": 1.0610822836711675,
|
| 189 |
+
"learning_rate": 4.4255319148936176e-05,
|
| 190 |
+
"loss": 0.7991,
|
| 191 |
+
"step": 26
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.285336856010568,
|
| 195 |
+
"grad_norm": 0.9499105971028553,
|
| 196 |
+
"learning_rate": 4.595744680851065e-05,
|
| 197 |
+
"loss": 0.7972,
|
| 198 |
+
"step": 27
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.29590488771466317,
|
| 202 |
+
"grad_norm": 1.7217798288469996,
|
| 203 |
+
"learning_rate": 4.765957446808511e-05,
|
| 204 |
+
"loss": 0.7937,
|
| 205 |
+
"step": 28
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.30647291941875826,
|
| 209 |
+
"grad_norm": 1.0574506769778116,
|
| 210 |
+
"learning_rate": 4.936170212765958e-05,
|
| 211 |
+
"loss": 0.7897,
|
| 212 |
+
"step": 29
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.31704095112285335,
|
| 216 |
+
"grad_norm": 1.9202057568924744,
|
| 217 |
+
"learning_rate": 5.106382978723405e-05,
|
| 218 |
+
"loss": 0.7991,
|
| 219 |
+
"step": 30
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.3276089828269485,
|
| 223 |
+
"grad_norm": 1.345605492975133,
|
| 224 |
+
"learning_rate": 5.276595744680851e-05,
|
| 225 |
+
"loss": 0.7846,
|
| 226 |
+
"step": 31
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.3381770145310436,
|
| 230 |
+
"grad_norm": 1.6956356026369743,
|
| 231 |
+
"learning_rate": 5.446808510638298e-05,
|
| 232 |
+
"loss": 0.791,
|
| 233 |
+
"step": 32
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.3487450462351387,
|
| 237 |
+
"grad_norm": 1.5678728230382326,
|
| 238 |
+
"learning_rate": 5.617021276595745e-05,
|
| 239 |
+
"loss": 0.7865,
|
| 240 |
+
"step": 33
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.35931307793923384,
|
| 244 |
+
"grad_norm": 1.1508014915876,
|
| 245 |
+
"learning_rate": 5.7872340425531925e-05,
|
| 246 |
+
"loss": 0.7796,
|
| 247 |
+
"step": 34
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.36988110964332893,
|
| 251 |
+
"grad_norm": 1.5510984064838533,
|
| 252 |
+
"learning_rate": 5.9574468085106384e-05,
|
| 253 |
+
"loss": 0.7771,
|
| 254 |
+
"step": 35
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.380449141347424,
|
| 258 |
+
"grad_norm": 1.400660302660845,
|
| 259 |
+
"learning_rate": 6.127659574468086e-05,
|
| 260 |
+
"loss": 0.7693,
|
| 261 |
+
"step": 36
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.3910171730515192,
|
| 265 |
+
"grad_norm": 1.390600461038752,
|
| 266 |
+
"learning_rate": 6.297872340425533e-05,
|
| 267 |
+
"loss": 0.7647,
|
| 268 |
+
"step": 37
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.40158520475561427,
|
| 272 |
+
"grad_norm": 1.3619693240026056,
|
| 273 |
+
"learning_rate": 6.468085106382979e-05,
|
| 274 |
+
"loss": 0.7704,
|
| 275 |
+
"step": 38
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.41215323645970936,
|
| 279 |
+
"grad_norm": 1.4593264335764384,
|
| 280 |
+
"learning_rate": 6.638297872340426e-05,
|
| 281 |
+
"loss": 0.7534,
|
| 282 |
+
"step": 39
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.4227212681638045,
|
| 286 |
+
"grad_norm": 1.3716746153165715,
|
| 287 |
+
"learning_rate": 6.808510638297873e-05,
|
| 288 |
+
"loss": 0.7614,
|
| 289 |
+
"step": 40
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.4332892998678996,
|
| 293 |
+
"grad_norm": 1.6638664908483813,
|
| 294 |
+
"learning_rate": 6.97872340425532e-05,
|
| 295 |
+
"loss": 0.7562,
|
| 296 |
+
"step": 41
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.4438573315719947,
|
| 300 |
+
"grad_norm": 1.1519315113645128,
|
| 301 |
+
"learning_rate": 7.148936170212766e-05,
|
| 302 |
+
"loss": 0.7493,
|
| 303 |
+
"step": 42
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.45442536327608984,
|
| 307 |
+
"grad_norm": 1.2801946128196509,
|
| 308 |
+
"learning_rate": 7.319148936170213e-05,
|
| 309 |
+
"loss": 0.7488,
|
| 310 |
+
"step": 43
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.46499339498018494,
|
| 314 |
+
"grad_norm": 63.9505110490317,
|
| 315 |
+
"learning_rate": 7.48936170212766e-05,
|
| 316 |
+
"loss": 0.7697,
|
| 317 |
+
"step": 44
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.47556142668428003,
|
| 321 |
+
"grad_norm": 2.589351765723454,
|
| 322 |
+
"learning_rate": 7.659574468085108e-05,
|
| 323 |
+
"loss": 0.7888,
|
| 324 |
+
"step": 45
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.4861294583883752,
|
| 328 |
+
"grad_norm": 1.302706513183853,
|
| 329 |
+
"learning_rate": 7.829787234042553e-05,
|
| 330 |
+
"loss": 0.7518,
|
| 331 |
+
"step": 46
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.4966974900924703,
|
| 335 |
+
"grad_norm": 1.968548255582277,
|
| 336 |
+
"learning_rate": 8e-05,
|
| 337 |
+
"loss": 0.771,
|
| 338 |
+
"step": 47
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.5072655217965654,
|
| 342 |
+
"grad_norm": 1.4317196073460219,
|
| 343 |
+
"learning_rate": 7.999889681839899e-05,
|
| 344 |
+
"loss": 0.7447,
|
| 345 |
+
"step": 48
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.5178335535006605,
|
| 349 |
+
"grad_norm": 1.4264803539125535,
|
| 350 |
+
"learning_rate": 7.999558733444641e-05,
|
| 351 |
+
"loss": 0.7495,
|
| 352 |
+
"step": 49
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.5284015852047557,
|
| 356 |
+
"grad_norm": 1.8250063568248835,
|
| 357 |
+
"learning_rate": 7.999007173069037e-05,
|
| 358 |
+
"loss": 0.7507,
|
| 359 |
+
"step": 50
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.5389696169088507,
|
| 363 |
+
"grad_norm": 32.42595517441195,
|
| 364 |
+
"learning_rate": 7.998235031136648e-05,
|
| 365 |
+
"loss": 0.8633,
|
| 366 |
+
"step": 51
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.5495376486129459,
|
| 370 |
+
"grad_norm": 2.457580825395341,
|
| 371 |
+
"learning_rate": 7.997242350238117e-05,
|
| 372 |
+
"loss": 0.7888,
|
| 373 |
+
"step": 52
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.560105680317041,
|
| 377 |
+
"grad_norm": 1.073949064797192,
|
| 378 |
+
"learning_rate": 7.996029185128804e-05,
|
| 379 |
+
"loss": 0.7334,
|
| 380 |
+
"step": 53
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.570673712021136,
|
| 384 |
+
"grad_norm": 1.7771625101062152,
|
| 385 |
+
"learning_rate": 7.994595602725781e-05,
|
| 386 |
+
"loss": 0.7529,
|
| 387 |
+
"step": 54
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.5812417437252312,
|
| 391 |
+
"grad_norm": 1.461447259519702,
|
| 392 |
+
"learning_rate": 7.992941682104139e-05,
|
| 393 |
+
"loss": 0.7563,
|
| 394 |
+
"step": 55
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.5918097754293263,
|
| 398 |
+
"grad_norm": 1.3789190576230315,
|
| 399 |
+
"learning_rate": 7.991067514492614e-05,
|
| 400 |
+
"loss": 0.7376,
|
| 401 |
+
"step": 56
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.6023778071334214,
|
| 405 |
+
"grad_norm": 1.0632278085259756,
|
| 406 |
+
"learning_rate": 7.988973203268567e-05,
|
| 407 |
+
"loss": 0.7418,
|
| 408 |
+
"step": 57
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.6129458388375165,
|
| 412 |
+
"grad_norm": 1.3753612784385743,
|
| 413 |
+
"learning_rate": 7.986658863952281e-05,
|
| 414 |
+
"loss": 0.7302,
|
| 415 |
+
"step": 58
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.6235138705416117,
|
| 419 |
+
"grad_norm": 1.3066179113984184,
|
| 420 |
+
"learning_rate": 7.984124624200583e-05,
|
| 421 |
+
"loss": 0.7397,
|
| 422 |
+
"step": 59
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.6340819022457067,
|
| 426 |
+
"grad_norm": 1.1484727063789566,
|
| 427 |
+
"learning_rate": 7.981370623799803e-05,
|
| 428 |
+
"loss": 0.7255,
|
| 429 |
+
"step": 60
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.6446499339498019,
|
| 433 |
+
"grad_norm": 0.992784588525566,
|
| 434 |
+
"learning_rate": 7.978397014658075e-05,
|
| 435 |
+
"loss": 0.7272,
|
| 436 |
+
"step": 61
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.655217965653897,
|
| 440 |
+
"grad_norm": 0.9709816709205683,
|
| 441 |
+
"learning_rate": 7.97520396079694e-05,
|
| 442 |
+
"loss": 0.7244,
|
| 443 |
+
"step": 62
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.665785997357992,
|
| 447 |
+
"grad_norm": 1.410223766064041,
|
| 448 |
+
"learning_rate": 7.971791638342313e-05,
|
| 449 |
+
"loss": 0.7277,
|
| 450 |
+
"step": 63
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.6763540290620872,
|
| 454 |
+
"grad_norm": 0.9859003082910416,
|
| 455 |
+
"learning_rate": 7.96816023551476e-05,
|
| 456 |
+
"loss": 0.7149,
|
| 457 |
+
"step": 64
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.6869220607661823,
|
| 461 |
+
"grad_norm": 1.0627873899695728,
|
| 462 |
+
"learning_rate": 7.964309952619121e-05,
|
| 463 |
+
"loss": 0.7076,
|
| 464 |
+
"step": 65
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.6974900924702774,
|
| 468 |
+
"grad_norm": 1.2987953031941966,
|
| 469 |
+
"learning_rate": 7.96024100203346e-05,
|
| 470 |
+
"loss": 0.7148,
|
| 471 |
+
"step": 66
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.7080581241743725,
|
| 475 |
+
"grad_norm": 0.9847502480003636,
|
| 476 |
+
"learning_rate": 7.955953608197345e-05,
|
| 477 |
+
"loss": 0.7192,
|
| 478 |
+
"step": 67
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.7186261558784677,
|
| 482 |
+
"grad_norm": 1.2534186109561034,
|
| 483 |
+
"learning_rate": 7.951448007599478e-05,
|
| 484 |
+
"loss": 0.7187,
|
| 485 |
+
"step": 68
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.7291941875825627,
|
| 489 |
+
"grad_norm": 1.1180727067911638,
|
| 490 |
+
"learning_rate": 7.946724448764644e-05,
|
| 491 |
+
"loss": 0.7093,
|
| 492 |
+
"step": 69
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.7397622192866579,
|
| 496 |
+
"grad_norm": 1.3232754557732016,
|
| 497 |
+
"learning_rate": 7.94178319224e-05,
|
| 498 |
+
"loss": 0.7124,
|
| 499 |
+
"step": 70
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.750330250990753,
|
| 503 |
+
"grad_norm": 0.9127404976454525,
|
| 504 |
+
"learning_rate": 7.936624510580712e-05,
|
| 505 |
+
"loss": 0.7075,
|
| 506 |
+
"step": 71
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.760898282694848,
|
| 510 |
+
"grad_norm": 0.769180480882775,
|
| 511 |
+
"learning_rate": 7.931248688334915e-05,
|
| 512 |
+
"loss": 0.6912,
|
| 513 |
+
"step": 72
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.7714663143989432,
|
| 517 |
+
"grad_norm": 0.8883585292926546,
|
| 518 |
+
"learning_rate": 7.925656022028017e-05,
|
| 519 |
+
"loss": 0.6987,
|
| 520 |
+
"step": 73
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.7820343461030383,
|
| 524 |
+
"grad_norm": 0.963883950564473,
|
| 525 |
+
"learning_rate": 7.919846820146348e-05,
|
| 526 |
+
"loss": 0.704,
|
| 527 |
+
"step": 74
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.7926023778071334,
|
| 531 |
+
"grad_norm": 0.8863497475275227,
|
| 532 |
+
"learning_rate": 7.913821403120139e-05,
|
| 533 |
+
"loss": 0.6945,
|
| 534 |
+
"step": 75
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.8031704095112285,
|
| 538 |
+
"grad_norm": 1.0426799292571676,
|
| 539 |
+
"learning_rate": 7.90758010330585e-05,
|
| 540 |
+
"loss": 0.6849,
|
| 541 |
+
"step": 76
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.8137384412153237,
|
| 545 |
+
"grad_norm": 0.8243890076961332,
|
| 546 |
+
"learning_rate": 7.901123264967836e-05,
|
| 547 |
+
"loss": 0.6979,
|
| 548 |
+
"step": 77
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.8243064729194187,
|
| 552 |
+
"grad_norm": 1.1652167740266006,
|
| 553 |
+
"learning_rate": 7.894451244259363e-05,
|
| 554 |
+
"loss": 0.6905,
|
| 555 |
+
"step": 78
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.8348745046235139,
|
| 559 |
+
"grad_norm": 1.2465660396622233,
|
| 560 |
+
"learning_rate": 7.887564409202953e-05,
|
| 561 |
+
"loss": 0.6872,
|
| 562 |
+
"step": 79
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.845442536327609,
|
| 566 |
+
"grad_norm": 0.5355690010674495,
|
| 567 |
+
"learning_rate": 7.880463139670091e-05,
|
| 568 |
+
"loss": 0.6848,
|
| 569 |
+
"step": 80
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.8560105680317041,
|
| 573 |
+
"grad_norm": 1.419104255807733,
|
| 574 |
+
"learning_rate": 7.873147827360273e-05,
|
| 575 |
+
"loss": 0.6878,
|
| 576 |
+
"step": 81
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.8665785997357992,
|
| 580 |
+
"grad_norm": 0.6749893210811997,
|
| 581 |
+
"learning_rate": 7.865618875779398e-05,
|
| 582 |
+
"loss": 0.678,
|
| 583 |
+
"step": 82
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.8771466314398944,
|
| 587 |
+
"grad_norm": 0.7732270985786504,
|
| 588 |
+
"learning_rate": 7.857876700217508e-05,
|
| 589 |
+
"loss": 0.6848,
|
| 590 |
+
"step": 83
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.8877146631439894,
|
| 594 |
+
"grad_norm": 0.6755156642812311,
|
| 595 |
+
"learning_rate": 7.849921727725882e-05,
|
| 596 |
+
"loss": 0.679,
|
| 597 |
+
"step": 84
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.8982826948480845,
|
| 601 |
+
"grad_norm": 0.6421189276197394,
|
| 602 |
+
"learning_rate": 7.841754397093487e-05,
|
| 603 |
+
"loss": 0.6728,
|
| 604 |
+
"step": 85
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.9088507265521797,
|
| 608 |
+
"grad_norm": 0.49153205280258144,
|
| 609 |
+
"learning_rate": 7.833375158822766e-05,
|
| 610 |
+
"loss": 0.6827,
|
| 611 |
+
"step": 86
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.9194187582562747,
|
| 615 |
+
"grad_norm": 0.7083774145182355,
|
| 616 |
+
"learning_rate": 7.824784475104795e-05,
|
| 617 |
+
"loss": 0.6813,
|
| 618 |
+
"step": 87
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.9299867899603699,
|
| 622 |
+
"grad_norm": 1.1312836778877742,
|
| 623 |
+
"learning_rate": 7.815982819793784e-05,
|
| 624 |
+
"loss": 0.6776,
|
| 625 |
+
"step": 88
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.940554821664465,
|
| 629 |
+
"grad_norm": 1.4566456596110349,
|
| 630 |
+
"learning_rate": 7.806970678380943e-05,
|
| 631 |
+
"loss": 0.6828,
|
| 632 |
+
"step": 89
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.9511228533685601,
|
| 636 |
+
"grad_norm": 0.641648090796948,
|
| 637 |
+
"learning_rate": 7.797748547967701e-05,
|
| 638 |
+
"loss": 0.6689,
|
| 639 |
+
"step": 90
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.9616908850726552,
|
| 643 |
+
"grad_norm": 1.2689027097316996,
|
| 644 |
+
"learning_rate": 7.788316937238287e-05,
|
| 645 |
+
"loss": 0.6887,
|
| 646 |
+
"step": 91
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.9722589167767504,
|
| 650 |
+
"grad_norm": 1.0288877116771369,
|
| 651 |
+
"learning_rate": 7.778676366431676e-05,
|
| 652 |
+
"loss": 0.6731,
|
| 653 |
+
"step": 92
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.9828269484808454,
|
| 657 |
+
"grad_norm": 1.111134713608855,
|
| 658 |
+
"learning_rate": 7.768827367312882e-05,
|
| 659 |
+
"loss": 0.6772,
|
| 660 |
+
"step": 93
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.9933949801849405,
|
| 664 |
+
"grad_norm": 0.9415602963246132,
|
| 665 |
+
"learning_rate": 7.758770483143634e-05,
|
| 666 |
+
"loss": 0.6764,
|
| 667 |
+
"step": 94
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 1.0039630118890357,
|
| 671 |
+
"grad_norm": 0.766369432032264,
|
| 672 |
+
"learning_rate": 7.748506268652415e-05,
|
| 673 |
+
"loss": 0.6581,
|
| 674 |
+
"step": 95
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 1.0145310435931307,
|
| 678 |
+
"grad_norm": 0.6235125904372976,
|
| 679 |
+
"learning_rate": 7.73803529000385e-05,
|
| 680 |
+
"loss": 0.639,
|
| 681 |
+
"step": 96
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 1.025099075297226,
|
| 685 |
+
"grad_norm": 0.5687108049822664,
|
| 686 |
+
"learning_rate": 7.727358124767491e-05,
|
| 687 |
+
"loss": 0.6307,
|
| 688 |
+
"step": 97
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 1.035667107001321,
|
| 692 |
+
"grad_norm": 0.7768949803155286,
|
| 693 |
+
"learning_rate": 7.71647536188595e-05,
|
| 694 |
+
"loss": 0.6287,
|
| 695 |
+
"step": 98
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 1.046235138705416,
|
| 699 |
+
"grad_norm": 1.0822134265138843,
|
| 700 |
+
"learning_rate": 7.705387601642416e-05,
|
| 701 |
+
"loss": 0.6316,
|
| 702 |
+
"step": 99
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 1.0568031704095113,
|
| 706 |
+
"grad_norm": 0.6600814825591903,
|
| 707 |
+
"learning_rate": 7.694095455627542e-05,
|
| 708 |
+
"loss": 0.6209,
|
| 709 |
+
"step": 100
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 1.0673712021136064,
|
| 713 |
+
"grad_norm": 0.5092321091554813,
|
| 714 |
+
"learning_rate": 7.682599546705716e-05,
|
| 715 |
+
"loss": 0.6268,
|
| 716 |
+
"step": 101
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 1.0779392338177014,
|
| 720 |
+
"grad_norm": 0.5538152767155732,
|
| 721 |
+
"learning_rate": 7.670900508980697e-05,
|
| 722 |
+
"loss": 0.629,
|
| 723 |
+
"step": 102
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 1.0885072655217967,
|
| 727 |
+
"grad_norm": 0.7759780022913368,
|
| 728 |
+
"learning_rate": 7.658998987760645e-05,
|
| 729 |
+
"loss": 0.6266,
|
| 730 |
+
"step": 103
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.0990752972258917,
|
| 734 |
+
"grad_norm": 1.0071580985859072,
|
| 735 |
+
"learning_rate": 7.646895639522518e-05,
|
| 736 |
+
"loss": 0.634,
|
| 737 |
+
"step": 104
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.1096433289299867,
|
| 741 |
+
"grad_norm": 0.9913392623454966,
|
| 742 |
+
"learning_rate": 7.634591131875875e-05,
|
| 743 |
+
"loss": 0.6334,
|
| 744 |
+
"step": 105
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.120211360634082,
|
| 748 |
+
"grad_norm": 0.9887888884996264,
|
| 749 |
+
"learning_rate": 7.622086143526036e-05,
|
| 750 |
+
"loss": 0.6271,
|
| 751 |
+
"step": 106
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.130779392338177,
|
| 755 |
+
"grad_norm": 1.0219278471051096,
|
| 756 |
+
"learning_rate": 7.609381364236655e-05,
|
| 757 |
+
"loss": 0.6224,
|
| 758 |
+
"step": 107
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.141347424042272,
|
| 762 |
+
"grad_norm": 0.9226284872759164,
|
| 763 |
+
"learning_rate": 7.59647749479167e-05,
|
| 764 |
+
"loss": 0.62,
|
| 765 |
+
"step": 108
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.1519154557463673,
|
| 769 |
+
"grad_norm": 0.9215854604914202,
|
| 770 |
+
"learning_rate": 7.583375246956648e-05,
|
| 771 |
+
"loss": 0.6295,
|
| 772 |
+
"step": 109
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.1624834874504624,
|
| 776 |
+
"grad_norm": 1.2986154846688018,
|
| 777 |
+
"learning_rate": 7.570075343439526e-05,
|
| 778 |
+
"loss": 0.6245,
|
| 779 |
+
"step": 110
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 1.1730515191545574,
|
| 783 |
+
"grad_norm": 0.5049907163605463,
|
| 784 |
+
"learning_rate": 7.556578517850747e-05,
|
| 785 |
+
"loss": 0.6244,
|
| 786 |
+
"step": 111
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 1.1836195508586527,
|
| 790 |
+
"grad_norm": 0.8014907412493998,
|
| 791 |
+
"learning_rate": 7.542885514662794e-05,
|
| 792 |
+
"loss": 0.6233,
|
| 793 |
+
"step": 112
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 1.1941875825627477,
|
| 797 |
+
"grad_norm": 1.1622767091260562,
|
| 798 |
+
"learning_rate": 7.528997089169128e-05,
|
| 799 |
+
"loss": 0.6235,
|
| 800 |
+
"step": 113
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 1.2047556142668427,
|
| 804 |
+
"grad_norm": 0.7380452282238854,
|
| 805 |
+
"learning_rate": 7.51491400744252e-05,
|
| 806 |
+
"loss": 0.6228,
|
| 807 |
+
"step": 114
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.215323645970938,
|
| 811 |
+
"grad_norm": 0.7597600873008943,
|
| 812 |
+
"learning_rate": 7.500637046292803e-05,
|
| 813 |
+
"loss": 0.621,
|
| 814 |
+
"step": 115
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 1.225891677675033,
|
| 818 |
+
"grad_norm": 0.7234884283262764,
|
| 819 |
+
"learning_rate": 7.48616699322402e-05,
|
| 820 |
+
"loss": 0.6194,
|
| 821 |
+
"step": 116
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 1.236459709379128,
|
| 825 |
+
"grad_norm": 0.49171596139374824,
|
| 826 |
+
"learning_rate": 7.471504646390987e-05,
|
| 827 |
+
"loss": 0.6197,
|
| 828 |
+
"step": 117
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 1.2470277410832233,
|
| 832 |
+
"grad_norm": 0.6952021893115369,
|
| 833 |
+
"learning_rate": 7.456650814555267e-05,
|
| 834 |
+
"loss": 0.6185,
|
| 835 |
+
"step": 118
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 1.2575957727873184,
|
| 839 |
+
"grad_norm": 0.7911960205209884,
|
| 840 |
+
"learning_rate": 7.441606317040558e-05,
|
| 841 |
+
"loss": 0.6204,
|
| 842 |
+
"step": 119
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.2681638044914134,
|
| 846 |
+
"grad_norm": 0.5604716814050085,
|
| 847 |
+
"learning_rate": 7.426371983687503e-05,
|
| 848 |
+
"loss": 0.6136,
|
| 849 |
+
"step": 120
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.2787318361955085,
|
| 853 |
+
"grad_norm": 0.5391019590519042,
|
| 854 |
+
"learning_rate": 7.410948654807916e-05,
|
| 855 |
+
"loss": 0.6141,
|
| 856 |
+
"step": 121
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 1.2892998678996037,
|
| 860 |
+
"grad_norm": 0.4473178756510967,
|
| 861 |
+
"learning_rate": 7.39533718113843e-05,
|
| 862 |
+
"loss": 0.618,
|
| 863 |
+
"step": 122
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 1.2998678996036988,
|
| 867 |
+
"grad_norm": 0.3676324142572552,
|
| 868 |
+
"learning_rate": 7.379538423793568e-05,
|
| 869 |
+
"loss": 0.6181,
|
| 870 |
+
"step": 123
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 1.310435931307794,
|
| 874 |
+
"grad_norm": 0.49641800089735266,
|
| 875 |
+
"learning_rate": 7.363553254218253e-05,
|
| 876 |
+
"loss": 0.6162,
|
| 877 |
+
"step": 124
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 1.321003963011889,
|
| 881 |
+
"grad_norm": 0.5596386402088841,
|
| 882 |
+
"learning_rate": 7.347382554139733e-05,
|
| 883 |
+
"loss": 0.6137,
|
| 884 |
+
"step": 125
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.331571994715984,
|
| 888 |
+
"grad_norm": 0.37385072844495554,
|
| 889 |
+
"learning_rate": 7.331027215518949e-05,
|
| 890 |
+
"loss": 0.6047,
|
| 891 |
+
"step": 126
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.3421400264200791,
|
| 895 |
+
"grad_norm": 0.46801024402125285,
|
| 896 |
+
"learning_rate": 7.31448814050133e-05,
|
| 897 |
+
"loss": 0.6074,
|
| 898 |
+
"step": 127
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.3527080581241744,
|
| 902 |
+
"grad_norm": 0.38199501822744514,
|
| 903 |
+
"learning_rate": 7.297766241367041e-05,
|
| 904 |
+
"loss": 0.6081,
|
| 905 |
+
"step": 128
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 1.3632760898282694,
|
| 909 |
+
"grad_norm": 0.2644636432614496,
|
| 910 |
+
"learning_rate": 7.280862440480658e-05,
|
| 911 |
+
"loss": 0.6083,
|
| 912 |
+
"step": 129
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.3738441215323647,
|
| 916 |
+
"grad_norm": 0.3642023310466234,
|
| 917 |
+
"learning_rate": 7.263777670240282e-05,
|
| 918 |
+
"loss": 0.6163,
|
| 919 |
+
"step": 130
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.3844121532364597,
|
| 923 |
+
"grad_norm": 0.3286254245460909,
|
| 924 |
+
"learning_rate": 7.246512873026125e-05,
|
| 925 |
+
"loss": 0.6105,
|
| 926 |
+
"step": 131
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.3949801849405548,
|
| 930 |
+
"grad_norm": 0.27645048682447476,
|
| 931 |
+
"learning_rate": 7.229069001148518e-05,
|
| 932 |
+
"loss": 0.6047,
|
| 933 |
+
"step": 132
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1.40554821664465,
|
| 937 |
+
"grad_norm": 0.31935620714929114,
|
| 938 |
+
"learning_rate": 7.211447016795388e-05,
|
| 939 |
+
"loss": 0.6159,
|
| 940 |
+
"step": 133
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.416116248348745,
|
| 944 |
+
"grad_norm": 0.27637648695223127,
|
| 945 |
+
"learning_rate": 7.193647891979177e-05,
|
| 946 |
+
"loss": 0.6076,
|
| 947 |
+
"step": 134
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.42668428005284,
|
| 951 |
+
"grad_norm": 0.2632383089776466,
|
| 952 |
+
"learning_rate": 7.17567260848324e-05,
|
| 953 |
+
"loss": 0.6134,
|
| 954 |
+
"step": 135
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.4372523117569354,
|
| 958 |
+
"grad_norm": 0.36576847146483216,
|
| 959 |
+
"learning_rate": 7.157522157807675e-05,
|
| 960 |
+
"loss": 0.6097,
|
| 961 |
+
"step": 136
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 1.4478203434610304,
|
| 965 |
+
"grad_norm": 0.49829210057276635,
|
| 966 |
+
"learning_rate": 7.139197541114645e-05,
|
| 967 |
+
"loss": 0.6076,
|
| 968 |
+
"step": 137
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 1.4583883751651254,
|
| 972 |
+
"grad_norm": 0.6715032042462543,
|
| 973 |
+
"learning_rate": 7.120699769173149e-05,
|
| 974 |
+
"loss": 0.6079,
|
| 975 |
+
"step": 138
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 1.4689564068692207,
|
| 979 |
+
"grad_norm": 0.8872767611634563,
|
| 980 |
+
"learning_rate": 7.10202986230327e-05,
|
| 981 |
+
"loss": 0.6106,
|
| 982 |
+
"step": 139
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 1.4795244385733157,
|
| 986 |
+
"grad_norm": 1.3094337587734477,
|
| 987 |
+
"learning_rate": 7.083188850319895e-05,
|
| 988 |
+
"loss": 0.6249,
|
| 989 |
+
"step": 140
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.4900924702774108,
|
| 993 |
+
"grad_norm": 0.5732521894924236,
|
| 994 |
+
"learning_rate": 7.064177772475912e-05,
|
| 995 |
+
"loss": 0.6184,
|
| 996 |
+
"step": 141
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.500660501981506,
|
| 1000 |
+
"grad_norm": 0.5968155749547341,
|
| 1001 |
+
"learning_rate": 7.044997677404888e-05,
|
| 1002 |
+
"loss": 0.6132,
|
| 1003 |
+
"step": 142
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 1.511228533685601,
|
| 1007 |
+
"grad_norm": 1.2108186694176202,
|
| 1008 |
+
"learning_rate": 7.025649623063223e-05,
|
| 1009 |
+
"loss": 0.6261,
|
| 1010 |
+
"step": 143
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.521796565389696,
|
| 1014 |
+
"grad_norm": 0.7492638271367434,
|
| 1015 |
+
"learning_rate": 7.006134676671791e-05,
|
| 1016 |
+
"loss": 0.6097,
|
| 1017 |
+
"step": 144
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 1.5323645970937911,
|
| 1021 |
+
"grad_norm": 0.520895839877394,
|
| 1022 |
+
"learning_rate": 6.986453914657083e-05,
|
| 1023 |
+
"loss": 0.6097,
|
| 1024 |
+
"step": 145
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 1.5429326287978864,
|
| 1028 |
+
"grad_norm": 0.6782854061408391,
|
| 1029 |
+
"learning_rate": 6.96660842259183e-05,
|
| 1030 |
+
"loss": 0.6175,
|
| 1031 |
+
"step": 146
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 1.5535006605019817,
|
| 1035 |
+
"grad_norm": 0.5917212461488441,
|
| 1036 |
+
"learning_rate": 6.946599295135116e-05,
|
| 1037 |
+
"loss": 0.6142,
|
| 1038 |
+
"step": 147
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 1.5640686922060767,
|
| 1042 |
+
"grad_norm": 0.5981371046939804,
|
| 1043 |
+
"learning_rate": 6.926427635972003e-05,
|
| 1044 |
+
"loss": 0.6083,
|
| 1045 |
+
"step": 148
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 1.5746367239101717,
|
| 1049 |
+
"grad_norm": 0.8111312440510982,
|
| 1050 |
+
"learning_rate": 6.906094557752654e-05,
|
| 1051 |
+
"loss": 0.6126,
|
| 1052 |
+
"step": 149
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 1.5852047556142668,
|
| 1056 |
+
"grad_norm": 0.9645422516897456,
|
| 1057 |
+
"learning_rate": 6.885601182030958e-05,
|
| 1058 |
+
"loss": 0.6143,
|
| 1059 |
+
"step": 150
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.5957727873183618,
|
| 1063 |
+
"grad_norm": 0.8593693771282125,
|
| 1064 |
+
"learning_rate": 6.864948639202667e-05,
|
| 1065 |
+
"loss": 0.6165,
|
| 1066 |
+
"step": 151
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.606340819022457,
|
| 1070 |
+
"grad_norm": 0.6782193407228015,
|
| 1071 |
+
"learning_rate": 6.844138068443043e-05,
|
| 1072 |
+
"loss": 0.617,
|
| 1073 |
+
"step": 152
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 1.6169088507265523,
|
| 1077 |
+
"grad_norm": 0.5133505641873833,
|
| 1078 |
+
"learning_rate": 6.823170617644029e-05,
|
| 1079 |
+
"loss": 0.6097,
|
| 1080 |
+
"step": 153
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 1.6274768824306474,
|
| 1084 |
+
"grad_norm": 0.3188527654318755,
|
| 1085 |
+
"learning_rate": 6.802047443350915e-05,
|
| 1086 |
+
"loss": 0.6019,
|
| 1087 |
+
"step": 154
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 1.6380449141347424,
|
| 1091 |
+
"grad_norm": 0.36046564912487544,
|
| 1092 |
+
"learning_rate": 6.78076971069857e-05,
|
| 1093 |
+
"loss": 0.603,
|
| 1094 |
+
"step": 155
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.6486129458388374,
|
| 1098 |
+
"grad_norm": 0.3434440344937455,
|
| 1099 |
+
"learning_rate": 6.759338593347148e-05,
|
| 1100 |
+
"loss": 0.614,
|
| 1101 |
+
"step": 156
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.6591809775429325,
|
| 1105 |
+
"grad_norm": 0.4207528199987266,
|
| 1106 |
+
"learning_rate": 6.737755273417367e-05,
|
| 1107 |
+
"loss": 0.6057,
|
| 1108 |
+
"step": 157
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 1.6697490092470277,
|
| 1112 |
+
"grad_norm": 0.4376375758003335,
|
| 1113 |
+
"learning_rate": 6.716020941425302e-05,
|
| 1114 |
+
"loss": 0.6101,
|
| 1115 |
+
"step": 158
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.680317040951123,
|
| 1119 |
+
"grad_norm": 0.3670986836675996,
|
| 1120 |
+
"learning_rate": 6.694136796216706e-05,
|
| 1121 |
+
"loss": 0.6074,
|
| 1122 |
+
"step": 159
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.690885072655218,
|
| 1126 |
+
"grad_norm": 0.3716710535382814,
|
| 1127 |
+
"learning_rate": 6.672104044900901e-05,
|
| 1128 |
+
"loss": 0.607,
|
| 1129 |
+
"step": 160
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.701453104359313,
|
| 1133 |
+
"grad_norm": 0.29349567743537014,
|
| 1134 |
+
"learning_rate": 6.649923902784178e-05,
|
| 1135 |
+
"loss": 0.6049,
|
| 1136 |
+
"step": 161
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.7120211360634081,
|
| 1140 |
+
"grad_norm": 0.37774756737109916,
|
| 1141 |
+
"learning_rate": 6.627597593302772e-05,
|
| 1142 |
+
"loss": 0.6004,
|
| 1143 |
+
"step": 162
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.7225891677675031,
|
| 1147 |
+
"grad_norm": 0.34520927764325005,
|
| 1148 |
+
"learning_rate": 6.605126347955376e-05,
|
| 1149 |
+
"loss": 0.5987,
|
| 1150 |
+
"step": 163
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 1.7331571994715984,
|
| 1154 |
+
"grad_norm": 0.37358546648640356,
|
| 1155 |
+
"learning_rate": 6.58251140623521e-05,
|
| 1156 |
+
"loss": 0.6037,
|
| 1157 |
+
"step": 164
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.7437252311756937,
|
| 1161 |
+
"grad_norm": 0.3803331681549159,
|
| 1162 |
+
"learning_rate": 6.559754015561655e-05,
|
| 1163 |
+
"loss": 0.6057,
|
| 1164 |
+
"step": 165
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.7542932628797887,
|
| 1168 |
+
"grad_norm": 0.419367642300496,
|
| 1169 |
+
"learning_rate": 6.536855431211445e-05,
|
| 1170 |
+
"loss": 0.6006,
|
| 1171 |
+
"step": 166
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.7648612945838837,
|
| 1175 |
+
"grad_norm": 0.3700144217552004,
|
| 1176 |
+
"learning_rate": 6.513816916249427e-05,
|
| 1177 |
+
"loss": 0.6029,
|
| 1178 |
+
"step": 167
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.7754293262879788,
|
| 1182 |
+
"grad_norm": 0.3304108780784769,
|
| 1183 |
+
"learning_rate": 6.490639741458891e-05,
|
| 1184 |
+
"loss": 0.609,
|
| 1185 |
+
"step": 168
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.7859973579920738,
|
| 1189 |
+
"grad_norm": 0.38491358241980217,
|
| 1190 |
+
"learning_rate": 6.46732518527148e-05,
|
| 1191 |
+
"loss": 0.5992,
|
| 1192 |
+
"step": 169
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 1.796565389696169,
|
| 1196 |
+
"grad_norm": 0.41792070367182416,
|
| 1197 |
+
"learning_rate": 6.443874533696662e-05,
|
| 1198 |
+
"loss": 0.5997,
|
| 1199 |
+
"step": 170
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.8071334214002643,
|
| 1203 |
+
"grad_norm": 0.31857578884550497,
|
| 1204 |
+
"learning_rate": 6.420289080250804e-05,
|
| 1205 |
+
"loss": 0.5976,
|
| 1206 |
+
"step": 171
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.8177014531043594,
|
| 1210 |
+
"grad_norm": 0.2921197050452141,
|
| 1211 |
+
"learning_rate": 6.396570125885823e-05,
|
| 1212 |
+
"loss": 0.6028,
|
| 1213 |
+
"step": 172
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.8282694848084544,
|
| 1217 |
+
"grad_norm": 0.32956079744279115,
|
| 1218 |
+
"learning_rate": 6.372718978917421e-05,
|
| 1219 |
+
"loss": 0.5959,
|
| 1220 |
+
"step": 173
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 1.8388375165125495,
|
| 1224 |
+
"grad_norm": 0.3363640406873935,
|
| 1225 |
+
"learning_rate": 6.348736954952923e-05,
|
| 1226 |
+
"loss": 0.5979,
|
| 1227 |
+
"step": 174
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.8494055482166445,
|
| 1231 |
+
"grad_norm": 0.2827997180555132,
|
| 1232 |
+
"learning_rate": 6.324625376818707e-05,
|
| 1233 |
+
"loss": 0.5963,
|
| 1234 |
+
"step": 175
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.8599735799207398,
|
| 1238 |
+
"grad_norm": 0.24952707846951472,
|
| 1239 |
+
"learning_rate": 6.300385574487242e-05,
|
| 1240 |
+
"loss": 0.5962,
|
| 1241 |
+
"step": 176
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.870541611624835,
|
| 1245 |
+
"grad_norm": 0.30585781576809906,
|
| 1246 |
+
"learning_rate": 6.276018885003727e-05,
|
| 1247 |
+
"loss": 0.5979,
|
| 1248 |
+
"step": 177
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.88110964332893,
|
| 1252 |
+
"grad_norm": 0.35383483624278156,
|
| 1253 |
+
"learning_rate": 6.251526652412335e-05,
|
| 1254 |
+
"loss": 0.5975,
|
| 1255 |
+
"step": 178
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.891677675033025,
|
| 1259 |
+
"grad_norm": 0.3057303385681727,
|
| 1260 |
+
"learning_rate": 6.226910227682087e-05,
|
| 1261 |
+
"loss": 0.5974,
|
| 1262 |
+
"step": 179
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.9022457067371201,
|
| 1266 |
+
"grad_norm": 0.3227218227112165,
|
| 1267 |
+
"learning_rate": 6.202170968632324e-05,
|
| 1268 |
+
"loss": 0.5984,
|
| 1269 |
+
"step": 180
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.9128137384412152,
|
| 1273 |
+
"grad_norm": 0.3099468501510001,
|
| 1274 |
+
"learning_rate": 6.177310239857815e-05,
|
| 1275 |
+
"loss": 0.6006,
|
| 1276 |
+
"step": 181
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 1.9233817701453104,
|
| 1280 |
+
"grad_norm": 0.3029754372937815,
|
| 1281 |
+
"learning_rate": 6.152329412653491e-05,
|
| 1282 |
+
"loss": 0.5937,
|
| 1283 |
+
"step": 182
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 1.9339498018494057,
|
| 1287 |
+
"grad_norm": 0.31003590660857305,
|
| 1288 |
+
"learning_rate": 6.127229864938798e-05,
|
| 1289 |
+
"loss": 0.6031,
|
| 1290 |
+
"step": 183
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.9445178335535007,
|
| 1294 |
+
"grad_norm": 0.4090438325379304,
|
| 1295 |
+
"learning_rate": 6.1020129811816985e-05,
|
| 1296 |
+
"loss": 0.5995,
|
| 1297 |
+
"step": 184
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.9550858652575958,
|
| 1301 |
+
"grad_norm": 0.47046399479064277,
|
| 1302 |
+
"learning_rate": 6.076680152322302e-05,
|
| 1303 |
+
"loss": 0.597,
|
| 1304 |
+
"step": 185
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 1.9656538969616908,
|
| 1308 |
+
"grad_norm": 0.4389393519962548,
|
| 1309 |
+
"learning_rate": 6.051232775696143e-05,
|
| 1310 |
+
"loss": 0.6003,
|
| 1311 |
+
"step": 186
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.9762219286657858,
|
| 1315 |
+
"grad_norm": 0.3946851961823208,
|
| 1316 |
+
"learning_rate": 6.025672254957106e-05,
|
| 1317 |
+
"loss": 0.5961,
|
| 1318 |
+
"step": 187
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 1.986789960369881,
|
| 1322 |
+
"grad_norm": 0.3581648113750019,
|
| 1323 |
+
"learning_rate": 6.000000000000001e-05,
|
| 1324 |
+
"loss": 0.6014,
|
| 1325 |
+
"step": 188
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 1.9973579920739764,
|
| 1329 |
+
"grad_norm": 0.32698044327640924,
|
| 1330 |
+
"learning_rate": 5.9742174268827936e-05,
|
| 1331 |
+
"loss": 0.6018,
|
| 1332 |
+
"step": 189
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 2.0079260237780714,
|
| 1336 |
+
"grad_norm": 0.3725595521245188,
|
| 1337 |
+
"learning_rate": 5.948325957748498e-05,
|
| 1338 |
+
"loss": 0.5527,
|
| 1339 |
+
"step": 190
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 2.0184940554821664,
|
| 1343 |
+
"grad_norm": 0.40702510391048585,
|
| 1344 |
+
"learning_rate": 5.9223270207467355e-05,
|
| 1345 |
+
"loss": 0.5457,
|
| 1346 |
+
"step": 191
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 2.0290620871862615,
|
| 1350 |
+
"grad_norm": 0.5137742231756651,
|
| 1351 |
+
"learning_rate": 5.896222049954951e-05,
|
| 1352 |
+
"loss": 0.5365,
|
| 1353 |
+
"step": 192
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 2.0396301188903565,
|
| 1357 |
+
"grad_norm": 0.5833504408833431,
|
| 1358 |
+
"learning_rate": 5.870012485299318e-05,
|
| 1359 |
+
"loss": 0.5339,
|
| 1360 |
+
"step": 193
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 2.050198150594452,
|
| 1364 |
+
"grad_norm": 0.5020674212672895,
|
| 1365 |
+
"learning_rate": 5.843699772475312e-05,
|
| 1366 |
+
"loss": 0.5344,
|
| 1367 |
+
"step": 194
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 2.060766182298547,
|
| 1371 |
+
"grad_norm": 0.5109669618772644,
|
| 1372 |
+
"learning_rate": 5.8172853628679676e-05,
|
| 1373 |
+
"loss": 0.5373,
|
| 1374 |
+
"step": 195
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 2.071334214002642,
|
| 1378 |
+
"grad_norm": 0.6079777296591362,
|
| 1379 |
+
"learning_rate": 5.790770713471816e-05,
|
| 1380 |
+
"loss": 0.5338,
|
| 1381 |
+
"step": 196
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 2.081902245706737,
|
| 1385 |
+
"grad_norm": 0.7717054202322884,
|
| 1386 |
+
"learning_rate": 5.764157286810527e-05,
|
| 1387 |
+
"loss": 0.5413,
|
| 1388 |
+
"step": 197
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 2.092470277410832,
|
| 1392 |
+
"grad_norm": 1.0263835129177117,
|
| 1393 |
+
"learning_rate": 5.7374465508562324e-05,
|
| 1394 |
+
"loss": 0.5419,
|
| 1395 |
+
"step": 198
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 2.103038309114927,
|
| 1399 |
+
"grad_norm": 1.1304188828796682,
|
| 1400 |
+
"learning_rate": 5.710639978948555e-05,
|
| 1401 |
+
"loss": 0.5401,
|
| 1402 |
+
"step": 199
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 2.1136063408190227,
|
| 1406 |
+
"grad_norm": 0.5561546704438505,
|
| 1407 |
+
"learning_rate": 5.6837390497133406e-05,
|
| 1408 |
+
"loss": 0.5371,
|
| 1409 |
+
"step": 200
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 2.1241743725231177,
|
| 1413 |
+
"grad_norm": 0.5094878513846206,
|
| 1414 |
+
"learning_rate": 5.6567452469810984e-05,
|
| 1415 |
+
"loss": 0.5307,
|
| 1416 |
+
"step": 201
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 2.1347424042272127,
|
| 1420 |
+
"grad_norm": 0.861314347361121,
|
| 1421 |
+
"learning_rate": 5.629660059705153e-05,
|
| 1422 |
+
"loss": 0.5405,
|
| 1423 |
+
"step": 202
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 2.1453104359313078,
|
| 1427 |
+
"grad_norm": 0.7744305400405808,
|
| 1428 |
+
"learning_rate": 5.602484981879519e-05,
|
| 1429 |
+
"loss": 0.5399,
|
| 1430 |
+
"step": 203
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 2.155878467635403,
|
| 1434 |
+
"grad_norm": 0.5568154422551339,
|
| 1435 |
+
"learning_rate": 5.5752215124564895e-05,
|
| 1436 |
+
"loss": 0.534,
|
| 1437 |
+
"step": 204
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 2.166446499339498,
|
| 1441 |
+
"grad_norm": 0.4932283903917851,
|
| 1442 |
+
"learning_rate": 5.547871155263955e-05,
|
| 1443 |
+
"loss": 0.5427,
|
| 1444 |
+
"step": 205
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 2.1770145310435933,
|
| 1448 |
+
"grad_norm": 0.6026007057555373,
|
| 1449 |
+
"learning_rate": 5.5204354189224596e-05,
|
| 1450 |
+
"loss": 0.5372,
|
| 1451 |
+
"step": 206
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 2.1875825627476884,
|
| 1455 |
+
"grad_norm": 0.5907730425638954,
|
| 1456 |
+
"learning_rate": 5.492915816761979e-05,
|
| 1457 |
+
"loss": 0.5339,
|
| 1458 |
+
"step": 207
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 2.1981505944517834,
|
| 1462 |
+
"grad_norm": 0.38702716198497655,
|
| 1463 |
+
"learning_rate": 5.465313866738454e-05,
|
| 1464 |
+
"loss": 0.5399,
|
| 1465 |
+
"step": 208
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 2.2087186261558784,
|
| 1469 |
+
"grad_norm": 0.4818332449688006,
|
| 1470 |
+
"learning_rate": 5.4376310913500514e-05,
|
| 1471 |
+
"loss": 0.5341,
|
| 1472 |
+
"step": 209
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 2.2192866578599735,
|
| 1476 |
+
"grad_norm": 0.48527023542347636,
|
| 1477 |
+
"learning_rate": 5.409869017553199e-05,
|
| 1478 |
+
"loss": 0.5443,
|
| 1479 |
+
"step": 210
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 2.2298546895640685,
|
| 1483 |
+
"grad_norm": 0.38988747831854015,
|
| 1484 |
+
"learning_rate": 5.382029176678345e-05,
|
| 1485 |
+
"loss": 0.5325,
|
| 1486 |
+
"step": 211
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 2.240422721268164,
|
| 1490 |
+
"grad_norm": 0.3382464323770342,
|
| 1491 |
+
"learning_rate": 5.354113104345503e-05,
|
| 1492 |
+
"loss": 0.5381,
|
| 1493 |
+
"step": 212
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 2.250990752972259,
|
| 1497 |
+
"grad_norm": 0.344596068112714,
|
| 1498 |
+
"learning_rate": 5.326122340379539e-05,
|
| 1499 |
+
"loss": 0.5393,
|
| 1500 |
+
"step": 213
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 2.261558784676354,
|
| 1504 |
+
"grad_norm": 0.38043335701597286,
|
| 1505 |
+
"learning_rate": 5.2980584287252456e-05,
|
| 1506 |
+
"loss": 0.5354,
|
| 1507 |
+
"step": 214
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 2.272126816380449,
|
| 1511 |
+
"grad_norm": 0.31330954952683393,
|
| 1512 |
+
"learning_rate": 5.269922917362171e-05,
|
| 1513 |
+
"loss": 0.5347,
|
| 1514 |
+
"step": 215
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 2.282694848084544,
|
| 1518 |
+
"grad_norm": 0.3394061867349534,
|
| 1519 |
+
"learning_rate": 5.241717358219239e-05,
|
| 1520 |
+
"loss": 0.5359,
|
| 1521 |
+
"step": 216
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 2.293262879788639,
|
| 1525 |
+
"grad_norm": 0.353132616874619,
|
| 1526 |
+
"learning_rate": 5.213443307089144e-05,
|
| 1527 |
+
"loss": 0.53,
|
| 1528 |
+
"step": 217
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 2.3038309114927347,
|
| 1532 |
+
"grad_norm": 0.2871048446747761,
|
| 1533 |
+
"learning_rate": 5.1851023235425366e-05,
|
| 1534 |
+
"loss": 0.5396,
|
| 1535 |
+
"step": 218
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 2.3143989431968297,
|
| 1539 |
+
"grad_norm": 0.3676997391938742,
|
| 1540 |
+
"learning_rate": 5.156695970841997e-05,
|
| 1541 |
+
"loss": 0.5319,
|
| 1542 |
+
"step": 219
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 2.3249669749009247,
|
| 1546 |
+
"grad_norm": 0.3034692647102483,
|
| 1547 |
+
"learning_rate": 5.128225815855805e-05,
|
| 1548 |
+
"loss": 0.539,
|
| 1549 |
+
"step": 220
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 2.33553500660502,
|
| 1553 |
+
"grad_norm": 0.29670953140630685,
|
| 1554 |
+
"learning_rate": 5.099693428971522e-05,
|
| 1555 |
+
"loss": 0.5357,
|
| 1556 |
+
"step": 221
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 2.346103038309115,
|
| 1560 |
+
"grad_norm": 0.3462570910330776,
|
| 1561 |
+
"learning_rate": 5.0711003840093583e-05,
|
| 1562 |
+
"loss": 0.5382,
|
| 1563 |
+
"step": 222
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 2.35667107001321,
|
| 1567 |
+
"grad_norm": 0.30320212822308745,
|
| 1568 |
+
"learning_rate": 5.042448258135371e-05,
|
| 1569 |
+
"loss": 0.5398,
|
| 1570 |
+
"step": 223
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 2.3672391017173053,
|
| 1574 |
+
"grad_norm": 0.3253569159099642,
|
| 1575 |
+
"learning_rate": 5.013738631774463e-05,
|
| 1576 |
+
"loss": 0.5403,
|
| 1577 |
+
"step": 224
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 2.3778071334214004,
|
| 1581 |
+
"grad_norm": 0.2992173696284864,
|
| 1582 |
+
"learning_rate": 4.984973088523216e-05,
|
| 1583 |
+
"loss": 0.5318,
|
| 1584 |
+
"step": 225
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 2.3883751651254954,
|
| 1588 |
+
"grad_norm": 0.23806839402955893,
|
| 1589 |
+
"learning_rate": 4.9561532150625305e-05,
|
| 1590 |
+
"loss": 0.5295,
|
| 1591 |
+
"step": 226
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 2.3989431968295905,
|
| 1595 |
+
"grad_norm": 0.2836158566149504,
|
| 1596 |
+
"learning_rate": 4.927280601070114e-05,
|
| 1597 |
+
"loss": 0.5273,
|
| 1598 |
+
"step": 227
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 2.4095112285336855,
|
| 1602 |
+
"grad_norm": 0.24042182902261217,
|
| 1603 |
+
"learning_rate": 4.898356839132793e-05,
|
| 1604 |
+
"loss": 0.5302,
|
| 1605 |
+
"step": 228
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 2.4200792602377805,
|
| 1609 |
+
"grad_norm": 0.23090066826959527,
|
| 1610 |
+
"learning_rate": 4.869383524658668e-05,
|
| 1611 |
+
"loss": 0.5378,
|
| 1612 |
+
"step": 229
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 2.430647291941876,
|
| 1616 |
+
"grad_norm": 0.23036421955638248,
|
| 1617 |
+
"learning_rate": 4.840362255789112e-05,
|
| 1618 |
+
"loss": 0.5324,
|
| 1619 |
+
"step": 230
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 2.441215323645971,
|
| 1623 |
+
"grad_norm": 0.2086043982624101,
|
| 1624 |
+
"learning_rate": 4.811294633310617e-05,
|
| 1625 |
+
"loss": 0.5355,
|
| 1626 |
+
"step": 231
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 2.451783355350066,
|
| 1630 |
+
"grad_norm": 0.2662049795738657,
|
| 1631 |
+
"learning_rate": 4.782182260566498e-05,
|
| 1632 |
+
"loss": 0.5321,
|
| 1633 |
+
"step": 232
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 2.462351387054161,
|
| 1637 |
+
"grad_norm": 0.17282690102361992,
|
| 1638 |
+
"learning_rate": 4.7530267433684546e-05,
|
| 1639 |
+
"loss": 0.5322,
|
| 1640 |
+
"step": 233
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 2.472919418758256,
|
| 1644 |
+
"grad_norm": 0.23839732691466814,
|
| 1645 |
+
"learning_rate": 4.723829689907993e-05,
|
| 1646 |
+
"loss": 0.5332,
|
| 1647 |
+
"step": 234
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 2.483487450462351,
|
| 1651 |
+
"grad_norm": 0.220755685419314,
|
| 1652 |
+
"learning_rate": 4.694592710667723e-05,
|
| 1653 |
+
"loss": 0.5289,
|
| 1654 |
+
"step": 235
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 2.4940554821664467,
|
| 1658 |
+
"grad_norm": 0.17887769697236458,
|
| 1659 |
+
"learning_rate": 4.665317418332521e-05,
|
| 1660 |
+
"loss": 0.5302,
|
| 1661 |
+
"step": 236
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 2.5046235138705417,
|
| 1665 |
+
"grad_norm": 0.21214814477629076,
|
| 1666 |
+
"learning_rate": 4.6360054277005826e-05,
|
| 1667 |
+
"loss": 0.5311,
|
| 1668 |
+
"step": 237
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 2.5151915455746368,
|
| 1672 |
+
"grad_norm": 0.169870987072969,
|
| 1673 |
+
"learning_rate": 4.606658355594344e-05,
|
| 1674 |
+
"loss": 0.5308,
|
| 1675 |
+
"step": 238
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 2.525759577278732,
|
| 1679 |
+
"grad_norm": 0.23328757071293307,
|
| 1680 |
+
"learning_rate": 4.577277820771307e-05,
|
| 1681 |
+
"loss": 0.531,
|
| 1682 |
+
"step": 239
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 2.536327608982827,
|
| 1686 |
+
"grad_norm": 0.2082393856633939,
|
| 1687 |
+
"learning_rate": 4.5478654438347414e-05,
|
| 1688 |
+
"loss": 0.5369,
|
| 1689 |
+
"step": 240
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 2.5468956406869223,
|
| 1693 |
+
"grad_norm": 0.14478055321484706,
|
| 1694 |
+
"learning_rate": 4.518422847144304e-05,
|
| 1695 |
+
"loss": 0.5323,
|
| 1696 |
+
"step": 241
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 2.557463672391017,
|
| 1700 |
+
"grad_norm": 0.22316005053591764,
|
| 1701 |
+
"learning_rate": 4.488951654726539e-05,
|
| 1702 |
+
"loss": 0.5286,
|
| 1703 |
+
"step": 242
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 2.5680317040951124,
|
| 1707 |
+
"grad_norm": 0.17285901709749715,
|
| 1708 |
+
"learning_rate": 4.4594534921853096e-05,
|
| 1709 |
+
"loss": 0.5362,
|
| 1710 |
+
"step": 243
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 2.5785997357992074,
|
| 1714 |
+
"grad_norm": 0.2122181023189799,
|
| 1715 |
+
"learning_rate": 4.429929986612125e-05,
|
| 1716 |
+
"loss": 0.5351,
|
| 1717 |
+
"step": 244
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 2.5891677675033025,
|
| 1721 |
+
"grad_norm": 0.19160763459008834,
|
| 1722 |
+
"learning_rate": 4.400382766496394e-05,
|
| 1723 |
+
"loss": 0.5322,
|
| 1724 |
+
"step": 245
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 2.5997357992073975,
|
| 1728 |
+
"grad_norm": 0.17931339221155976,
|
| 1729 |
+
"learning_rate": 4.3708134616355934e-05,
|
| 1730 |
+
"loss": 0.5291,
|
| 1731 |
+
"step": 246
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 2.6103038309114925,
|
| 1735 |
+
"grad_norm": 0.18662561546217404,
|
| 1736 |
+
"learning_rate": 4.341223703045379e-05,
|
| 1737 |
+
"loss": 0.5348,
|
| 1738 |
+
"step": 247
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 2.620871862615588,
|
| 1742 |
+
"grad_norm": 0.15830313964721893,
|
| 1743 |
+
"learning_rate": 4.311615122869613e-05,
|
| 1744 |
+
"loss": 0.5434,
|
| 1745 |
+
"step": 248
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 2.631439894319683,
|
| 1749 |
+
"grad_norm": 0.19402467010937804,
|
| 1750 |
+
"learning_rate": 4.281989354290341e-05,
|
| 1751 |
+
"loss": 0.5433,
|
| 1752 |
+
"step": 249
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 2.642007926023778,
|
| 1756 |
+
"grad_norm": 0.20042343107408042,
|
| 1757 |
+
"learning_rate": 4.2523480314376996e-05,
|
| 1758 |
+
"loss": 0.5327,
|
| 1759 |
+
"step": 250
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 2.652575957727873,
|
| 1763 |
+
"grad_norm": 0.165259104609407,
|
| 1764 |
+
"learning_rate": 4.222692789299794e-05,
|
| 1765 |
+
"loss": 0.5389,
|
| 1766 |
+
"step": 251
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 2.663143989431968,
|
| 1770 |
+
"grad_norm": 0.21602254674705257,
|
| 1771 |
+
"learning_rate": 4.193025263632495e-05,
|
| 1772 |
+
"loss": 0.5254,
|
| 1773 |
+
"step": 252
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 2.6737120211360637,
|
| 1777 |
+
"grad_norm": 0.1679859643732669,
|
| 1778 |
+
"learning_rate": 4.163347090869227e-05,
|
| 1779 |
+
"loss": 0.5375,
|
| 1780 |
+
"step": 253
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 2.6842800528401582,
|
| 1784 |
+
"grad_norm": 0.17225237105164173,
|
| 1785 |
+
"learning_rate": 4.133659908030699e-05,
|
| 1786 |
+
"loss": 0.5342,
|
| 1787 |
+
"step": 254
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 2.6948480845442537,
|
| 1791 |
+
"grad_norm": 0.1620251583234328,
|
| 1792 |
+
"learning_rate": 4.103965352634604e-05,
|
| 1793 |
+
"loss": 0.5328,
|
| 1794 |
+
"step": 255
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 2.7054161162483488,
|
| 1798 |
+
"grad_norm": 0.16680355622763784,
|
| 1799 |
+
"learning_rate": 4.0742650626053004e-05,
|
| 1800 |
+
"loss": 0.5293,
|
| 1801 |
+
"step": 256
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 2.715984147952444,
|
| 1805 |
+
"grad_norm": 0.15521980569419397,
|
| 1806 |
+
"learning_rate": 4.044560676183462e-05,
|
| 1807 |
+
"loss": 0.5371,
|
| 1808 |
+
"step": 257
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 2.726552179656539,
|
| 1812 |
+
"grad_norm": 0.1574325532000935,
|
| 1813 |
+
"learning_rate": 4.014853831835721e-05,
|
| 1814 |
+
"loss": 0.5331,
|
| 1815 |
+
"step": 258
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 2.737120211360634,
|
| 1819 |
+
"grad_norm": 0.16871282749949595,
|
| 1820 |
+
"learning_rate": 3.985146168164281e-05,
|
| 1821 |
+
"loss": 0.5388,
|
| 1822 |
+
"step": 259
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 2.7476882430647294,
|
| 1826 |
+
"grad_norm": 0.1553640449532965,
|
| 1827 |
+
"learning_rate": 3.9554393238165386e-05,
|
| 1828 |
+
"loss": 0.5284,
|
| 1829 |
+
"step": 260
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 2.7582562747688244,
|
| 1833 |
+
"grad_norm": 0.15753872116807552,
|
| 1834 |
+
"learning_rate": 3.9257349373947016e-05,
|
| 1835 |
+
"loss": 0.5342,
|
| 1836 |
+
"step": 261
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 2.7688243064729194,
|
| 1840 |
+
"grad_norm": 0.1510488137537568,
|
| 1841 |
+
"learning_rate": 3.896034647365398e-05,
|
| 1842 |
+
"loss": 0.5329,
|
| 1843 |
+
"step": 262
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 2.7793923381770145,
|
| 1847 |
+
"grad_norm": 0.17598454072482386,
|
| 1848 |
+
"learning_rate": 3.866340091969303e-05,
|
| 1849 |
+
"loss": 0.5344,
|
| 1850 |
+
"step": 263
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 2.7899603698811095,
|
| 1854 |
+
"grad_norm": 0.16623830505420029,
|
| 1855 |
+
"learning_rate": 3.836652909130774e-05,
|
| 1856 |
+
"loss": 0.5273,
|
| 1857 |
+
"step": 264
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 2.800528401585205,
|
| 1861 |
+
"grad_norm": 0.1959528022905974,
|
| 1862 |
+
"learning_rate": 3.806974736367507e-05,
|
| 1863 |
+
"loss": 0.5324,
|
| 1864 |
+
"step": 265
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 2.8110964332893,
|
| 1868 |
+
"grad_norm": 0.14859823867573574,
|
| 1869 |
+
"learning_rate": 3.7773072107002084e-05,
|
| 1870 |
+
"loss": 0.5334,
|
| 1871 |
+
"step": 266
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 2.821664464993395,
|
| 1875 |
+
"grad_norm": 0.19136907727546654,
|
| 1876 |
+
"learning_rate": 3.747651968562302e-05,
|
| 1877 |
+
"loss": 0.5283,
|
| 1878 |
+
"step": 267
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 2.83223249669749,
|
| 1882 |
+
"grad_norm": 0.1741838181803494,
|
| 1883 |
+
"learning_rate": 3.718010645709661e-05,
|
| 1884 |
+
"loss": 0.5374,
|
| 1885 |
+
"step": 268
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 2.842800528401585,
|
| 1889 |
+
"grad_norm": 0.15429083348412015,
|
| 1890 |
+
"learning_rate": 3.688384877130388e-05,
|
| 1891 |
+
"loss": 0.5348,
|
| 1892 |
+
"step": 269
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 2.85336856010568,
|
| 1896 |
+
"grad_norm": 0.19719371528147198,
|
| 1897 |
+
"learning_rate": 3.658776296954622e-05,
|
| 1898 |
+
"loss": 0.5339,
|
| 1899 |
+
"step": 270
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 2.8639365918097752,
|
| 1903 |
+
"grad_norm": 0.15789471821578627,
|
| 1904 |
+
"learning_rate": 3.629186538364408e-05,
|
| 1905 |
+
"loss": 0.5339,
|
| 1906 |
+
"step": 271
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 2.8745046235138707,
|
| 1910 |
+
"grad_norm": 0.16444809265187393,
|
| 1911 |
+
"learning_rate": 3.5996172335036065e-05,
|
| 1912 |
+
"loss": 0.54,
|
| 1913 |
+
"step": 272
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 2.8850726552179657,
|
| 1917 |
+
"grad_norm": 0.18205663719249998,
|
| 1918 |
+
"learning_rate": 3.570070013387876e-05,
|
| 1919 |
+
"loss": 0.5326,
|
| 1920 |
+
"step": 273
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 2.895640686922061,
|
| 1924 |
+
"grad_norm": 0.1638276415587385,
|
| 1925 |
+
"learning_rate": 3.540546507814692e-05,
|
| 1926 |
+
"loss": 0.5336,
|
| 1927 |
+
"step": 274
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 2.906208718626156,
|
| 1931 |
+
"grad_norm": 0.14430321990734493,
|
| 1932 |
+
"learning_rate": 3.5110483452734633e-05,
|
| 1933 |
+
"loss": 0.5333,
|
| 1934 |
+
"step": 275
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 2.916776750330251,
|
| 1938 |
+
"grad_norm": 0.15779067389210275,
|
| 1939 |
+
"learning_rate": 3.4815771528556976e-05,
|
| 1940 |
+
"loss": 0.5295,
|
| 1941 |
+
"step": 276
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 2.9273447820343463,
|
| 1945 |
+
"grad_norm": 0.16510325621420335,
|
| 1946 |
+
"learning_rate": 3.452134556165259e-05,
|
| 1947 |
+
"loss": 0.5322,
|
| 1948 |
+
"step": 277
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 2.9379128137384414,
|
| 1952 |
+
"grad_norm": 0.15335282480316584,
|
| 1953 |
+
"learning_rate": 3.4227221792286945e-05,
|
| 1954 |
+
"loss": 0.5266,
|
| 1955 |
+
"step": 278
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 2.9484808454425364,
|
| 1959 |
+
"grad_norm": 0.14207500884610488,
|
| 1960 |
+
"learning_rate": 3.393341644405657e-05,
|
| 1961 |
+
"loss": 0.5284,
|
| 1962 |
+
"step": 279
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 2.9590488771466315,
|
| 1966 |
+
"grad_norm": 0.14817401425997304,
|
| 1967 |
+
"learning_rate": 3.363994572299418e-05,
|
| 1968 |
+
"loss": 0.5305,
|
| 1969 |
+
"step": 280
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 2.9696169088507265,
|
| 1973 |
+
"grad_norm": 0.13417873579347814,
|
| 1974 |
+
"learning_rate": 3.3346825816674804e-05,
|
| 1975 |
+
"loss": 0.5285,
|
| 1976 |
+
"step": 281
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 2.9801849405548215,
|
| 1980 |
+
"grad_norm": 0.1933271122465297,
|
| 1981 |
+
"learning_rate": 3.305407289332279e-05,
|
| 1982 |
+
"loss": 0.5336,
|
| 1983 |
+
"step": 282
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 2.9907529722589166,
|
| 1987 |
+
"grad_norm": 0.14082276570620816,
|
| 1988 |
+
"learning_rate": 3.276170310092008e-05,
|
| 1989 |
+
"loss": 0.5314,
|
| 1990 |
+
"step": 283
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 3.001321003963012,
|
| 1994 |
+
"grad_norm": 0.16572738564748216,
|
| 1995 |
+
"learning_rate": 3.246973256631546e-05,
|
| 1996 |
+
"loss": 0.5233,
|
| 1997 |
+
"step": 284
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 3.011889035667107,
|
| 2001 |
+
"grad_norm": 0.27880171253028246,
|
| 2002 |
+
"learning_rate": 3.217817739433502e-05,
|
| 2003 |
+
"loss": 0.4778,
|
| 2004 |
+
"step": 285
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 3.022457067371202,
|
| 2008 |
+
"grad_norm": 0.2862871579036325,
|
| 2009 |
+
"learning_rate": 3.1887053666893834e-05,
|
| 2010 |
+
"loss": 0.4793,
|
| 2011 |
+
"step": 286
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 3.033025099075297,
|
| 2015 |
+
"grad_norm": 0.25228138426980523,
|
| 2016 |
+
"learning_rate": 3.159637744210888e-05,
|
| 2017 |
+
"loss": 0.4754,
|
| 2018 |
+
"step": 287
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 3.043593130779392,
|
| 2022 |
+
"grad_norm": 0.2079690373301947,
|
| 2023 |
+
"learning_rate": 3.130616475341332e-05,
|
| 2024 |
+
"loss": 0.4723,
|
| 2025 |
+
"step": 288
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 3.0541611624834872,
|
| 2029 |
+
"grad_norm": 0.18869207117374362,
|
| 2030 |
+
"learning_rate": 3.101643160867208e-05,
|
| 2031 |
+
"loss": 0.4762,
|
| 2032 |
+
"step": 289
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 3.0647291941875827,
|
| 2036 |
+
"grad_norm": 0.22869432322360056,
|
| 2037 |
+
"learning_rate": 3.072719398929887e-05,
|
| 2038 |
+
"loss": 0.4737,
|
| 2039 |
+
"step": 290
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 3.0752972258916778,
|
| 2043 |
+
"grad_norm": 0.15890007803087935,
|
| 2044 |
+
"learning_rate": 3.0438467849374702e-05,
|
| 2045 |
+
"loss": 0.473,
|
| 2046 |
+
"step": 291
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 3.085865257595773,
|
| 2050 |
+
"grad_norm": 0.210513738135415,
|
| 2051 |
+
"learning_rate": 3.0150269114767862e-05,
|
| 2052 |
+
"loss": 0.4767,
|
| 2053 |
+
"step": 292
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 3.096433289299868,
|
| 2057 |
+
"grad_norm": 0.18412299124477924,
|
| 2058 |
+
"learning_rate": 2.9862613682255383e-05,
|
| 2059 |
+
"loss": 0.4761,
|
| 2060 |
+
"step": 293
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 3.107001321003963,
|
| 2064 |
+
"grad_norm": 0.16086763510358643,
|
| 2065 |
+
"learning_rate": 2.957551741864631e-05,
|
| 2066 |
+
"loss": 0.4743,
|
| 2067 |
+
"step": 294
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 3.117569352708058,
|
| 2071 |
+
"grad_norm": 0.1836629217850037,
|
| 2072 |
+
"learning_rate": 2.928899615990643e-05,
|
| 2073 |
+
"loss": 0.475,
|
| 2074 |
+
"step": 295
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 3.1281373844121534,
|
| 2078 |
+
"grad_norm": 0.16351418203553833,
|
| 2079 |
+
"learning_rate": 2.90030657102848e-05,
|
| 2080 |
+
"loss": 0.478,
|
| 2081 |
+
"step": 296
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 3.1387054161162484,
|
| 2085 |
+
"grad_norm": 0.1543546082591788,
|
| 2086 |
+
"learning_rate": 2.8717741841441964e-05,
|
| 2087 |
+
"loss": 0.4736,
|
| 2088 |
+
"step": 297
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 3.1492734478203435,
|
| 2092 |
+
"grad_norm": 0.14811662444817492,
|
| 2093 |
+
"learning_rate": 2.8433040291580053e-05,
|
| 2094 |
+
"loss": 0.4814,
|
| 2095 |
+
"step": 298
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 3.1598414795244385,
|
| 2099 |
+
"grad_norm": 0.15161900774346382,
|
| 2100 |
+
"learning_rate": 2.8148976764574648e-05,
|
| 2101 |
+
"loss": 0.4723,
|
| 2102 |
+
"step": 299
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 3.1704095112285335,
|
| 2106 |
+
"grad_norm": 0.14342849635394805,
|
| 2107 |
+
"learning_rate": 2.7865566929108573e-05,
|
| 2108 |
+
"loss": 0.4761,
|
| 2109 |
+
"step": 300
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 3.180977542932629,
|
| 2113 |
+
"grad_norm": 0.1660043564516897,
|
| 2114 |
+
"learning_rate": 2.758282641780762e-05,
|
| 2115 |
+
"loss": 0.4797,
|
| 2116 |
+
"step": 301
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 3.191545574636724,
|
| 2120 |
+
"grad_norm": 0.141920732705004,
|
| 2121 |
+
"learning_rate": 2.7300770826378302e-05,
|
| 2122 |
+
"loss": 0.4782,
|
| 2123 |
+
"step": 302
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 3.202113606340819,
|
| 2127 |
+
"grad_norm": 0.14950725725564776,
|
| 2128 |
+
"learning_rate": 2.7019415712747558e-05,
|
| 2129 |
+
"loss": 0.4748,
|
| 2130 |
+
"step": 303
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 3.212681638044914,
|
| 2134 |
+
"grad_norm": 0.14208430168695735,
|
| 2135 |
+
"learning_rate": 2.6738776596204624e-05,
|
| 2136 |
+
"loss": 0.4748,
|
| 2137 |
+
"step": 304
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 3.223249669749009,
|
| 2141 |
+
"grad_norm": 0.14961017547836333,
|
| 2142 |
+
"learning_rate": 2.6458868956544984e-05,
|
| 2143 |
+
"loss": 0.4803,
|
| 2144 |
+
"step": 305
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 3.233817701453104,
|
| 2148 |
+
"grad_norm": 0.15645442679040064,
|
| 2149 |
+
"learning_rate": 2.6179708233216557e-05,
|
| 2150 |
+
"loss": 0.4758,
|
| 2151 |
+
"step": 306
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 3.2443857331571992,
|
| 2155 |
+
"grad_norm": 0.1537770287650884,
|
| 2156 |
+
"learning_rate": 2.590130982446802e-05,
|
| 2157 |
+
"loss": 0.4737,
|
| 2158 |
+
"step": 307
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 3.2549537648612947,
|
| 2162 |
+
"grad_norm": 0.14713329423087662,
|
| 2163 |
+
"learning_rate": 2.5623689086499496e-05,
|
| 2164 |
+
"loss": 0.4733,
|
| 2165 |
+
"step": 308
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 3.2655217965653898,
|
| 2169 |
+
"grad_norm": 0.15291508052862732,
|
| 2170 |
+
"learning_rate": 2.5346861332615476e-05,
|
| 2171 |
+
"loss": 0.4782,
|
| 2172 |
+
"step": 309
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 3.276089828269485,
|
| 2176 |
+
"grad_norm": 0.13920928762276993,
|
| 2177 |
+
"learning_rate": 2.5070841832380212e-05,
|
| 2178 |
+
"loss": 0.4709,
|
| 2179 |
+
"step": 310
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 3.28665785997358,
|
| 2183 |
+
"grad_norm": 0.1465474915994889,
|
| 2184 |
+
"learning_rate": 2.4795645810775414e-05,
|
| 2185 |
+
"loss": 0.4826,
|
| 2186 |
+
"step": 311
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 3.297225891677675,
|
| 2190 |
+
"grad_norm": 0.1316301584217805,
|
| 2191 |
+
"learning_rate": 2.4521288447360457e-05,
|
| 2192 |
+
"loss": 0.4731,
|
| 2193 |
+
"step": 312
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 3.3077939233817704,
|
| 2197 |
+
"grad_norm": 0.1461801975123496,
|
| 2198 |
+
"learning_rate": 2.424778487543512e-05,
|
| 2199 |
+
"loss": 0.4734,
|
| 2200 |
+
"step": 313
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 3.3183619550858654,
|
| 2204 |
+
"grad_norm": 0.12719699667003487,
|
| 2205 |
+
"learning_rate": 2.3975150181204817e-05,
|
| 2206 |
+
"loss": 0.4737,
|
| 2207 |
+
"step": 314
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 3.3289299867899604,
|
| 2211 |
+
"grad_norm": 0.13764467132837127,
|
| 2212 |
+
"learning_rate": 2.370339940294848e-05,
|
| 2213 |
+
"loss": 0.4703,
|
| 2214 |
+
"step": 315
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 3.3394980184940555,
|
| 2218 |
+
"grad_norm": 0.12510614551548813,
|
| 2219 |
+
"learning_rate": 2.3432547530189033e-05,
|
| 2220 |
+
"loss": 0.4786,
|
| 2221 |
+
"step": 316
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 3.3500660501981505,
|
| 2225 |
+
"grad_norm": 0.13577816786562447,
|
| 2226 |
+
"learning_rate": 2.316260950286661e-05,
|
| 2227 |
+
"loss": 0.4775,
|
| 2228 |
+
"step": 317
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 3.3606340819022456,
|
| 2232 |
+
"grad_norm": 0.13089413394472932,
|
| 2233 |
+
"learning_rate": 2.2893600210514464e-05,
|
| 2234 |
+
"loss": 0.4756,
|
| 2235 |
+
"step": 318
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 3.3712021136063406,
|
| 2239 |
+
"grad_norm": 0.1333288679621952,
|
| 2240 |
+
"learning_rate": 2.2625534491437672e-05,
|
| 2241 |
+
"loss": 0.4811,
|
| 2242 |
+
"step": 319
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 3.381770145310436,
|
| 2246 |
+
"grad_norm": 0.13920771603982868,
|
| 2247 |
+
"learning_rate": 2.2358427131894732e-05,
|
| 2248 |
+
"loss": 0.4815,
|
| 2249 |
+
"step": 320
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 3.392338177014531,
|
| 2253 |
+
"grad_norm": 0.12578897335151773,
|
| 2254 |
+
"learning_rate": 2.2092292865281845e-05,
|
| 2255 |
+
"loss": 0.477,
|
| 2256 |
+
"step": 321
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 3.402906208718626,
|
| 2260 |
+
"grad_norm": 0.1406551312359615,
|
| 2261 |
+
"learning_rate": 2.1827146371320334e-05,
|
| 2262 |
+
"loss": 0.4761,
|
| 2263 |
+
"step": 322
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 3.413474240422721,
|
| 2267 |
+
"grad_norm": 0.11847800398887365,
|
| 2268 |
+
"learning_rate": 2.156300227524688e-05,
|
| 2269 |
+
"loss": 0.4794,
|
| 2270 |
+
"step": 323
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 3.4240422721268162,
|
| 2274 |
+
"grad_norm": 0.1289493089958156,
|
| 2275 |
+
"learning_rate": 2.1299875147006838e-05,
|
| 2276 |
+
"loss": 0.4803,
|
| 2277 |
+
"step": 324
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 3.4346103038309117,
|
| 2281 |
+
"grad_norm": 0.12509059453118254,
|
| 2282 |
+
"learning_rate": 2.10377795004505e-05,
|
| 2283 |
+
"loss": 0.4774,
|
| 2284 |
+
"step": 325
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 3.4451783355350067,
|
| 2288 |
+
"grad_norm": 0.13580138707758002,
|
| 2289 |
+
"learning_rate": 2.0776729792532652e-05,
|
| 2290 |
+
"loss": 0.4733,
|
| 2291 |
+
"step": 326
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 3.455746367239102,
|
| 2295 |
+
"grad_norm": 0.12405647314750252,
|
| 2296 |
+
"learning_rate": 2.0516740422515022e-05,
|
| 2297 |
+
"loss": 0.4762,
|
| 2298 |
+
"step": 327
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 3.466314398943197,
|
| 2302 |
+
"grad_norm": 0.14580809381579368,
|
| 2303 |
+
"learning_rate": 2.0257825731172077e-05,
|
| 2304 |
+
"loss": 0.4764,
|
| 2305 |
+
"step": 328
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 3.476882430647292,
|
| 2309 |
+
"grad_norm": 0.1281590923924517,
|
| 2310 |
+
"learning_rate": 2.0000000000000012e-05,
|
| 2311 |
+
"loss": 0.4706,
|
| 2312 |
+
"step": 329
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 3.487450462351387,
|
| 2316 |
+
"grad_norm": 0.1298579667890172,
|
| 2317 |
+
"learning_rate": 1.9743277450428962e-05,
|
| 2318 |
+
"loss": 0.4725,
|
| 2319 |
+
"step": 330
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 3.498018494055482,
|
| 2323 |
+
"grad_norm": 0.13901143297798416,
|
| 2324 |
+
"learning_rate": 1.9487672243038594e-05,
|
| 2325 |
+
"loss": 0.4811,
|
| 2326 |
+
"step": 331
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 3.5085865257595774,
|
| 2330 |
+
"grad_norm": 0.12006662533988818,
|
| 2331 |
+
"learning_rate": 1.9233198476777003e-05,
|
| 2332 |
+
"loss": 0.4755,
|
| 2333 |
+
"step": 332
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 3.5191545574636725,
|
| 2337 |
+
"grad_norm": 0.13509800282835904,
|
| 2338 |
+
"learning_rate": 1.897987018818302e-05,
|
| 2339 |
+
"loss": 0.4736,
|
| 2340 |
+
"step": 333
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 3.5297225891677675,
|
| 2344 |
+
"grad_norm": 0.12372407695325513,
|
| 2345 |
+
"learning_rate": 1.8727701350612026e-05,
|
| 2346 |
+
"loss": 0.4771,
|
| 2347 |
+
"step": 334
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 3.5402906208718625,
|
| 2351 |
+
"grad_norm": 0.13277097757447007,
|
| 2352 |
+
"learning_rate": 1.8476705873465097e-05,
|
| 2353 |
+
"loss": 0.477,
|
| 2354 |
+
"step": 335
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 3.5508586525759576,
|
| 2358 |
+
"grad_norm": 0.1190584597907601,
|
| 2359 |
+
"learning_rate": 1.8226897601421858e-05,
|
| 2360 |
+
"loss": 0.472,
|
| 2361 |
+
"step": 336
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 3.561426684280053,
|
| 2365 |
+
"grad_norm": 0.13360122007454706,
|
| 2366 |
+
"learning_rate": 1.7978290313676774e-05,
|
| 2367 |
+
"loss": 0.4812,
|
| 2368 |
+
"step": 337
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 3.571994715984148,
|
| 2372 |
+
"grad_norm": 0.12353280522144573,
|
| 2373 |
+
"learning_rate": 1.7730897723179144e-05,
|
| 2374 |
+
"loss": 0.4735,
|
| 2375 |
+
"step": 338
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 3.582562747688243,
|
| 2379 |
+
"grad_norm": 0.12197612268234642,
|
| 2380 |
+
"learning_rate": 1.748473347587666e-05,
|
| 2381 |
+
"loss": 0.4758,
|
| 2382 |
+
"step": 339
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 3.593130779392338,
|
| 2386 |
+
"grad_norm": 0.13321021249065187,
|
| 2387 |
+
"learning_rate": 1.7239811149962756e-05,
|
| 2388 |
+
"loss": 0.4777,
|
| 2389 |
+
"step": 340
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 3.603698811096433,
|
| 2393 |
+
"grad_norm": 0.11696601002999722,
|
| 2394 |
+
"learning_rate": 1.6996144255127586e-05,
|
| 2395 |
+
"loss": 0.473,
|
| 2396 |
+
"step": 341
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 3.6142668428005287,
|
| 2400 |
+
"grad_norm": 0.12376625663933627,
|
| 2401 |
+
"learning_rate": 1.675374623181294e-05,
|
| 2402 |
+
"loss": 0.475,
|
| 2403 |
+
"step": 342
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 3.6248348745046233,
|
| 2407 |
+
"grad_norm": 0.11118193452823608,
|
| 2408 |
+
"learning_rate": 1.6512630450470784e-05,
|
| 2409 |
+
"loss": 0.4679,
|
| 2410 |
+
"step": 343
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 3.6354029062087188,
|
| 2414 |
+
"grad_norm": 0.1194452215053386,
|
| 2415 |
+
"learning_rate": 1.6272810210825794e-05,
|
| 2416 |
+
"loss": 0.4748,
|
| 2417 |
+
"step": 344
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 3.645970937912814,
|
| 2421 |
+
"grad_norm": 0.12011227175979526,
|
| 2422 |
+
"learning_rate": 1.6034298741141768e-05,
|
| 2423 |
+
"loss": 0.4766,
|
| 2424 |
+
"step": 345
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 3.656538969616909,
|
| 2428 |
+
"grad_norm": 0.10988264863047191,
|
| 2429 |
+
"learning_rate": 1.579710919749196e-05,
|
| 2430 |
+
"loss": 0.4735,
|
| 2431 |
+
"step": 346
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 3.667107001321004,
|
| 2435 |
+
"grad_norm": 0.11947936011899343,
|
| 2436 |
+
"learning_rate": 1.5561254663033393e-05,
|
| 2437 |
+
"loss": 0.4801,
|
| 2438 |
+
"step": 347
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 3.677675033025099,
|
| 2442 |
+
"grad_norm": 0.11387001376651971,
|
| 2443 |
+
"learning_rate": 1.53267481472852e-05,
|
| 2444 |
+
"loss": 0.4761,
|
| 2445 |
+
"step": 348
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 3.6882430647291944,
|
| 2449 |
+
"grad_norm": 0.10828606028620547,
|
| 2450 |
+
"learning_rate": 1.5093602585411078e-05,
|
| 2451 |
+
"loss": 0.4745,
|
| 2452 |
+
"step": 349
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 3.6988110964332894,
|
| 2456 |
+
"grad_norm": 0.1158137481167655,
|
| 2457 |
+
"learning_rate": 1.4861830837505733e-05,
|
| 2458 |
+
"loss": 0.4739,
|
| 2459 |
+
"step": 350
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 3.7093791281373845,
|
| 2463 |
+
"grad_norm": 0.11883853717597995,
|
| 2464 |
+
"learning_rate": 1.4631445687885553e-05,
|
| 2465 |
+
"loss": 0.479,
|
| 2466 |
+
"step": 351
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 3.7199471598414795,
|
| 2470 |
+
"grad_norm": 0.10711524384366743,
|
| 2471 |
+
"learning_rate": 1.4402459844383451e-05,
|
| 2472 |
+
"loss": 0.4751,
|
| 2473 |
+
"step": 352
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 3.7305151915455745,
|
| 2477 |
+
"grad_norm": 0.1179009653082842,
|
| 2478 |
+
"learning_rate": 1.4174885937647905e-05,
|
| 2479 |
+
"loss": 0.4782,
|
| 2480 |
+
"step": 353
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 3.74108322324967,
|
| 2484 |
+
"grad_norm": 0.11366521654185796,
|
| 2485 |
+
"learning_rate": 1.3948736520446246e-05,
|
| 2486 |
+
"loss": 0.4727,
|
| 2487 |
+
"step": 354
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 3.7516512549537646,
|
| 2491 |
+
"grad_norm": 0.10920060902225881,
|
| 2492 |
+
"learning_rate": 1.372402406697229e-05,
|
| 2493 |
+
"loss": 0.4725,
|
| 2494 |
+
"step": 355
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 3.76221928665786,
|
| 2498 |
+
"grad_norm": 0.12313189801852109,
|
| 2499 |
+
"learning_rate": 1.3500760972158223e-05,
|
| 2500 |
+
"loss": 0.4809,
|
| 2501 |
+
"step": 356
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 3.772787318361955,
|
| 2505 |
+
"grad_norm": 0.11049761112743338,
|
| 2506 |
+
"learning_rate": 1.3278959550991011e-05,
|
| 2507 |
+
"loss": 0.4782,
|
| 2508 |
+
"step": 357
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 3.78335535006605,
|
| 2512 |
+
"grad_norm": 0.10430827010366775,
|
| 2513 |
+
"learning_rate": 1.3058632037832957e-05,
|
| 2514 |
+
"loss": 0.4743,
|
| 2515 |
+
"step": 358
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 3.793923381770145,
|
| 2519 |
+
"grad_norm": 0.11402898251067686,
|
| 2520 |
+
"learning_rate": 1.2839790585747008e-05,
|
| 2521 |
+
"loss": 0.4767,
|
| 2522 |
+
"step": 359
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 3.8044914134742402,
|
| 2526 |
+
"grad_norm": 0.11162325433379153,
|
| 2527 |
+
"learning_rate": 1.2622447265826345e-05,
|
| 2528 |
+
"loss": 0.4812,
|
| 2529 |
+
"step": 360
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 3.8150594451783357,
|
| 2533 |
+
"grad_norm": 0.10511190972273543,
|
| 2534 |
+
"learning_rate": 1.2406614066528543e-05,
|
| 2535 |
+
"loss": 0.4768,
|
| 2536 |
+
"step": 361
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 3.8256274768824308,
|
| 2540 |
+
"grad_norm": 0.10681643814143399,
|
| 2541 |
+
"learning_rate": 1.219230289301431e-05,
|
| 2542 |
+
"loss": 0.4717,
|
| 2543 |
+
"step": 362
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 3.836195508586526,
|
| 2547 |
+
"grad_norm": 0.11345833004338865,
|
| 2548 |
+
"learning_rate": 1.1979525566490845e-05,
|
| 2549 |
+
"loss": 0.4694,
|
| 2550 |
+
"step": 363
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 3.846763540290621,
|
| 2554 |
+
"grad_norm": 0.10629326422923135,
|
| 2555 |
+
"learning_rate": 1.176829382355973e-05,
|
| 2556 |
+
"loss": 0.475,
|
| 2557 |
+
"step": 364
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 3.857331571994716,
|
| 2561 |
+
"grad_norm": 0.10487140068250281,
|
| 2562 |
+
"learning_rate": 1.1558619315569572e-05,
|
| 2563 |
+
"loss": 0.4727,
|
| 2564 |
+
"step": 365
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 3.8678996036988114,
|
| 2568 |
+
"grad_norm": 0.10353148274468497,
|
| 2569 |
+
"learning_rate": 1.1350513607973351e-05,
|
| 2570 |
+
"loss": 0.4749,
|
| 2571 |
+
"step": 366
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 3.878467635402906,
|
| 2575 |
+
"grad_norm": 0.11029487277113502,
|
| 2576 |
+
"learning_rate": 1.1143988179690441e-05,
|
| 2577 |
+
"loss": 0.4756,
|
| 2578 |
+
"step": 367
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 3.8890356671070014,
|
| 2582 |
+
"grad_norm": 0.10444430423294958,
|
| 2583 |
+
"learning_rate": 1.093905442247348e-05,
|
| 2584 |
+
"loss": 0.4715,
|
| 2585 |
+
"step": 368
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 3.8996036988110965,
|
| 2589 |
+
"grad_norm": 0.10279747300542211,
|
| 2590 |
+
"learning_rate": 1.073572364027999e-05,
|
| 2591 |
+
"loss": 0.4701,
|
| 2592 |
+
"step": 369
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 3.9101717305151915,
|
| 2596 |
+
"grad_norm": 0.10814918538118634,
|
| 2597 |
+
"learning_rate": 1.0534007048648846e-05,
|
| 2598 |
+
"loss": 0.4773,
|
| 2599 |
+
"step": 370
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 3.9207397622192866,
|
| 2603 |
+
"grad_norm": 0.10568724216385697,
|
| 2604 |
+
"learning_rate": 1.0333915774081698e-05,
|
| 2605 |
+
"loss": 0.4742,
|
| 2606 |
+
"step": 371
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 3.9313077939233816,
|
| 2610 |
+
"grad_norm": 0.10446279448267286,
|
| 2611 |
+
"learning_rate": 1.0135460853429166e-05,
|
| 2612 |
+
"loss": 0.4703,
|
| 2613 |
+
"step": 372
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 3.941875825627477,
|
| 2617 |
+
"grad_norm": 0.10539141636554224,
|
| 2618 |
+
"learning_rate": 9.938653233282105e-06,
|
| 2619 |
+
"loss": 0.4788,
|
| 2620 |
+
"step": 373
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 3.952443857331572,
|
| 2624 |
+
"grad_norm": 0.10404420155804553,
|
| 2625 |
+
"learning_rate": 9.74350376936779e-06,
|
| 2626 |
+
"loss": 0.4766,
|
| 2627 |
+
"step": 374
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 3.963011889035667,
|
| 2631 |
+
"grad_norm": 0.11291514851116513,
|
| 2632 |
+
"learning_rate": 9.550023225951124e-06,
|
| 2633 |
+
"loss": 0.4786,
|
| 2634 |
+
"step": 375
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 3.973579920739762,
|
| 2638 |
+
"grad_norm": 0.11054229268613286,
|
| 2639 |
+
"learning_rate": 9.358222275240884e-06,
|
| 2640 |
+
"loss": 0.4715,
|
| 2641 |
+
"step": 376
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 3.984147952443857,
|
| 2645 |
+
"grad_norm": 0.10047846272456959,
|
| 2646 |
+
"learning_rate": 9.168111496801071e-06,
|
| 2647 |
+
"loss": 0.4767,
|
| 2648 |
+
"step": 377
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 3.9947159841479527,
|
| 2652 |
+
"grad_norm": 0.10427667029654042,
|
| 2653 |
+
"learning_rate": 8.979701376967313e-06,
|
| 2654 |
+
"loss": 0.472,
|
| 2655 |
+
"step": 378
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 4.005284015852047,
|
| 2659 |
+
"grad_norm": 0.15160742471081606,
|
| 2660 |
+
"learning_rate": 8.793002308268521e-06,
|
| 2661 |
+
"loss": 0.4599,
|
| 2662 |
+
"step": 379
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 4.015852047556143,
|
| 2666 |
+
"grad_norm": 0.18761042591080226,
|
| 2667 |
+
"learning_rate": 8.608024588853561e-06,
|
| 2668 |
+
"loss": 0.4482,
|
| 2669 |
+
"step": 380
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 4.026420079260237,
|
| 2673 |
+
"grad_norm": 0.12647969828019093,
|
| 2674 |
+
"learning_rate": 8.424778421923258e-06,
|
| 2675 |
+
"loss": 0.4377,
|
| 2676 |
+
"step": 381
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 4.036988110964333,
|
| 2680 |
+
"grad_norm": 0.1707261073721691,
|
| 2681 |
+
"learning_rate": 8.24327391516761e-06,
|
| 2682 |
+
"loss": 0.4438,
|
| 2683 |
+
"step": 382
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 4.047556142668428,
|
| 2687 |
+
"grad_norm": 0.20005656576433853,
|
| 2688 |
+
"learning_rate": 8.06352108020823e-06,
|
| 2689 |
+
"loss": 0.4427,
|
| 2690 |
+
"step": 383
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 4.058124174372523,
|
| 2694 |
+
"grad_norm": 0.14945783232518764,
|
| 2695 |
+
"learning_rate": 7.885529832046134e-06,
|
| 2696 |
+
"loss": 0.4404,
|
| 2697 |
+
"step": 384
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 4.068692206076618,
|
| 2701 |
+
"grad_norm": 0.13331270252131155,
|
| 2702 |
+
"learning_rate": 7.709309988514824e-06,
|
| 2703 |
+
"loss": 0.438,
|
| 2704 |
+
"step": 385
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 4.079260237780713,
|
| 2708 |
+
"grad_norm": 0.1373177545653382,
|
| 2709 |
+
"learning_rate": 7.534871269738753e-06,
|
| 2710 |
+
"loss": 0.442,
|
| 2711 |
+
"step": 386
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 4.0898282694848085,
|
| 2715 |
+
"grad_norm": 0.14428406658074794,
|
| 2716 |
+
"learning_rate": 7.362223297597184e-06,
|
| 2717 |
+
"loss": 0.444,
|
| 2718 |
+
"step": 387
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 4.100396301188904,
|
| 2722 |
+
"grad_norm": 0.1403209142453575,
|
| 2723 |
+
"learning_rate": 7.191375595193433e-06,
|
| 2724 |
+
"loss": 0.4423,
|
| 2725 |
+
"step": 388
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 4.110964332892999,
|
| 2729 |
+
"grad_norm": 0.12198908241232029,
|
| 2730 |
+
"learning_rate": 7.022337586329597e-06,
|
| 2731 |
+
"loss": 0.4436,
|
| 2732 |
+
"step": 389
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 4.121532364597094,
|
| 2736 |
+
"grad_norm": 0.11000767534504179,
|
| 2737 |
+
"learning_rate": 6.855118594986718e-06,
|
| 2738 |
+
"loss": 0.4443,
|
| 2739 |
+
"step": 390
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 4.132100396301189,
|
| 2743 |
+
"grad_norm": 0.1243765198168875,
|
| 2744 |
+
"learning_rate": 6.6897278448105405e-06,
|
| 2745 |
+
"loss": 0.4389,
|
| 2746 |
+
"step": 391
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 4.142668428005284,
|
| 2750 |
+
"grad_norm": 0.1326492943126363,
|
| 2751 |
+
"learning_rate": 6.526174458602681e-06,
|
| 2752 |
+
"loss": 0.4461,
|
| 2753 |
+
"step": 392
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 4.153236459709379,
|
| 2757 |
+
"grad_norm": 0.11686776812880073,
|
| 2758 |
+
"learning_rate": 6.364467457817482e-06,
|
| 2759 |
+
"loss": 0.4384,
|
| 2760 |
+
"step": 393
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 4.163804491413474,
|
| 2764 |
+
"grad_norm": 0.10054607780091933,
|
| 2765 |
+
"learning_rate": 6.20461576206433e-06,
|
| 2766 |
+
"loss": 0.4389,
|
| 2767 |
+
"step": 394
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 4.17437252311757,
|
| 2771 |
+
"grad_norm": 0.11593044593000827,
|
| 2772 |
+
"learning_rate": 6.046628188615718e-06,
|
| 2773 |
+
"loss": 0.4398,
|
| 2774 |
+
"step": 395
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 4.184940554821664,
|
| 2778 |
+
"grad_norm": 0.11430820998845044,
|
| 2779 |
+
"learning_rate": 5.890513451920843e-06,
|
| 2780 |
+
"loss": 0.4367,
|
| 2781 |
+
"step": 396
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 4.19550858652576,
|
| 2785 |
+
"grad_norm": 0.11287261939767183,
|
| 2786 |
+
"learning_rate": 5.736280163124974e-06,
|
| 2787 |
+
"loss": 0.4444,
|
| 2788 |
+
"step": 397
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 4.206076618229854,
|
| 2792 |
+
"grad_norm": 0.10321211605129811,
|
| 2793 |
+
"learning_rate": 5.583936829594434e-06,
|
| 2794 |
+
"loss": 0.4436,
|
| 2795 |
+
"step": 398
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 4.21664464993395,
|
| 2799 |
+
"grad_norm": 0.10361621115480053,
|
| 2800 |
+
"learning_rate": 5.4334918544473436e-06,
|
| 2801 |
+
"loss": 0.4405,
|
| 2802 |
+
"step": 399
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 4.227212681638045,
|
| 2806 |
+
"grad_norm": 0.10863502177296244,
|
| 2807 |
+
"learning_rate": 5.284953536090131e-06,
|
| 2808 |
+
"loss": 0.443,
|
| 2809 |
+
"step": 400
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 4.23778071334214,
|
| 2813 |
+
"grad_norm": 0.10536497157476482,
|
| 2814 |
+
"learning_rate": 5.1383300677598024e-06,
|
| 2815 |
+
"loss": 0.4497,
|
| 2816 |
+
"step": 401
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 4.248348745046235,
|
| 2820 |
+
"grad_norm": 0.10402891881587309,
|
| 2821 |
+
"learning_rate": 4.993629537071978e-06,
|
| 2822 |
+
"loss": 0.4396,
|
| 2823 |
+
"step": 402
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 4.25891677675033,
|
| 2827 |
+
"grad_norm": 0.10369536487473348,
|
| 2828 |
+
"learning_rate": 4.850859925574809e-06,
|
| 2829 |
+
"loss": 0.4393,
|
| 2830 |
+
"step": 403
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 4.2694848084544255,
|
| 2834 |
+
"grad_norm": 0.10111203263697728,
|
| 2835 |
+
"learning_rate": 4.710029108308733e-06,
|
| 2836 |
+
"loss": 0.441,
|
| 2837 |
+
"step": 404
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 4.28005284015852,
|
| 2841 |
+
"grad_norm": 0.09912151235410817,
|
| 2842 |
+
"learning_rate": 4.571144853372063e-06,
|
| 2843 |
+
"loss": 0.4403,
|
| 2844 |
+
"step": 405
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 4.2906208718626155,
|
| 2848 |
+
"grad_norm": 0.09757157331199569,
|
| 2849 |
+
"learning_rate": 4.434214821492542e-06,
|
| 2850 |
+
"loss": 0.4473,
|
| 2851 |
+
"step": 406
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 4.301188903566711,
|
| 2855 |
+
"grad_norm": 0.1002102756021715,
|
| 2856 |
+
"learning_rate": 4.299246565604756e-06,
|
| 2857 |
+
"loss": 0.4383,
|
| 2858 |
+
"step": 407
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 4.311756935270806,
|
| 2862 |
+
"grad_norm": 0.10119754294494109,
|
| 2863 |
+
"learning_rate": 4.166247530433531e-06,
|
| 2864 |
+
"loss": 0.4363,
|
| 2865 |
+
"step": 408
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 4.322324966974901,
|
| 2869 |
+
"grad_norm": 0.10100539004557267,
|
| 2870 |
+
"learning_rate": 4.035225052083309e-06,
|
| 2871 |
+
"loss": 0.4406,
|
| 2872 |
+
"step": 409
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 4.332892998678996,
|
| 2876 |
+
"grad_norm": 0.09823672032855243,
|
| 2877 |
+
"learning_rate": 3.906186357633455e-06,
|
| 2878 |
+
"loss": 0.4432,
|
| 2879 |
+
"step": 410
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 4.343461030383091,
|
| 2883 |
+
"grad_norm": 0.09520265410670203,
|
| 2884 |
+
"learning_rate": 3.779138564739646e-06,
|
| 2885 |
+
"loss": 0.436,
|
| 2886 |
+
"step": 411
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 4.354029062087187,
|
| 2890 |
+
"grad_norm": 0.10024695357191145,
|
| 2891 |
+
"learning_rate": 3.6540886812412547e-06,
|
| 2892 |
+
"loss": 0.4405,
|
| 2893 |
+
"step": 412
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 4.364597093791281,
|
| 2897 |
+
"grad_norm": 0.10079785013841713,
|
| 2898 |
+
"learning_rate": 3.5310436047748263e-06,
|
| 2899 |
+
"loss": 0.4403,
|
| 2900 |
+
"step": 413
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 4.375165125495377,
|
| 2904 |
+
"grad_norm": 0.09516762618496769,
|
| 2905 |
+
"learning_rate": 3.4100101223935743e-06,
|
| 2906 |
+
"loss": 0.4408,
|
| 2907 |
+
"step": 414
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 4.385733157199471,
|
| 2911 |
+
"grad_norm": 0.09404930151960895,
|
| 2912 |
+
"learning_rate": 3.290994910193037e-06,
|
| 2913 |
+
"loss": 0.4442,
|
| 2914 |
+
"step": 415
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 4.396301188903567,
|
| 2918 |
+
"grad_norm": 0.09531190586303208,
|
| 2919 |
+
"learning_rate": 3.174004532942845e-06,
|
| 2920 |
+
"loss": 0.4451,
|
| 2921 |
+
"step": 416
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 4.406869220607662,
|
| 2925 |
+
"grad_norm": 0.0951071973632443,
|
| 2926 |
+
"learning_rate": 3.059045443724582e-06,
|
| 2927 |
+
"loss": 0.4423,
|
| 2928 |
+
"step": 417
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 4.417437252311757,
|
| 2932 |
+
"grad_norm": 0.09705103462701054,
|
| 2933 |
+
"learning_rate": 2.946123983575846e-06,
|
| 2934 |
+
"loss": 0.4372,
|
| 2935 |
+
"step": 418
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 4.428005284015852,
|
| 2939 |
+
"grad_norm": 0.09365563642961511,
|
| 2940 |
+
"learning_rate": 2.8352463811404952e-06,
|
| 2941 |
+
"loss": 0.4397,
|
| 2942 |
+
"step": 419
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 4.438573315719947,
|
| 2946 |
+
"grad_norm": 0.09694934512919057,
|
| 2947 |
+
"learning_rate": 2.726418752325084e-06,
|
| 2948 |
+
"loss": 0.4465,
|
| 2949 |
+
"step": 420
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 4.449141347424042,
|
| 2953 |
+
"grad_norm": 0.09546394710630239,
|
| 2954 |
+
"learning_rate": 2.619647099961502e-06,
|
| 2955 |
+
"loss": 0.4457,
|
| 2956 |
+
"step": 421
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 4.459709379128137,
|
| 2960 |
+
"grad_norm": 0.09255730663301787,
|
| 2961 |
+
"learning_rate": 2.514937313475865e-06,
|
| 2962 |
+
"loss": 0.4405,
|
| 2963 |
+
"step": 422
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 4.4702774108322325,
|
| 2967 |
+
"grad_norm": 0.09338235711048254,
|
| 2968 |
+
"learning_rate": 2.4122951685636674e-06,
|
| 2969 |
+
"loss": 0.4395,
|
| 2970 |
+
"step": 423
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 4.480845442536328,
|
| 2974 |
+
"grad_norm": 0.09435019171855656,
|
| 2975 |
+
"learning_rate": 2.3117263268712e-06,
|
| 2976 |
+
"loss": 0.44,
|
| 2977 |
+
"step": 424
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 4.491413474240423,
|
| 2981 |
+
"grad_norm": 0.09413409031470825,
|
| 2982 |
+
"learning_rate": 2.2132363356832532e-06,
|
| 2983 |
+
"loss": 0.4432,
|
| 2984 |
+
"step": 425
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 4.501981505944518,
|
| 2988 |
+
"grad_norm": 0.09468855592447012,
|
| 2989 |
+
"learning_rate": 2.11683062761713e-06,
|
| 2990 |
+
"loss": 0.4422,
|
| 2991 |
+
"step": 426
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 4.512549537648613,
|
| 2995 |
+
"grad_norm": 0.09055553100177934,
|
| 2996 |
+
"learning_rate": 2.0225145203230044e-06,
|
| 2997 |
+
"loss": 0.4401,
|
| 2998 |
+
"step": 427
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 4.523117569352708,
|
| 3002 |
+
"grad_norm": 0.09082079754983816,
|
| 3003 |
+
"learning_rate": 1.930293216190586e-06,
|
| 3004 |
+
"loss": 0.4367,
|
| 3005 |
+
"step": 428
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 4.533685601056803,
|
| 3009 |
+
"grad_norm": 0.09114595588995444,
|
| 3010 |
+
"learning_rate": 1.8401718020621694e-06,
|
| 3011 |
+
"loss": 0.4383,
|
| 3012 |
+
"step": 429
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 4.544253632760898,
|
| 3016 |
+
"grad_norm": 0.09147630992219807,
|
| 3017 |
+
"learning_rate": 1.7521552489520566e-06,
|
| 3018 |
+
"loss": 0.4455,
|
| 3019 |
+
"step": 430
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 4.554821664464994,
|
| 3023 |
+
"grad_norm": 0.09327852060770747,
|
| 3024 |
+
"learning_rate": 1.666248411772342e-06,
|
| 3025 |
+
"loss": 0.441,
|
| 3026 |
+
"step": 431
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 4.565389696169088,
|
| 3030 |
+
"grad_norm": 0.09307772965424148,
|
| 3031 |
+
"learning_rate": 1.5824560290651404e-06,
|
| 3032 |
+
"loss": 0.444,
|
| 3033 |
+
"step": 432
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 4.575957727873184,
|
| 3037 |
+
"grad_norm": 0.09415356635848661,
|
| 3038 |
+
"learning_rate": 1.5007827227411942e-06,
|
| 3039 |
+
"loss": 0.4417,
|
| 3040 |
+
"step": 433
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 4.586525759577278,
|
| 3044 |
+
"grad_norm": 0.09176517213906957,
|
| 3045 |
+
"learning_rate": 1.4212329978249417e-06,
|
| 3046 |
+
"loss": 0.4425,
|
| 3047 |
+
"step": 434
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 4.597093791281374,
|
| 3051 |
+
"grad_norm": 0.09036761344520468,
|
| 3052 |
+
"learning_rate": 1.3438112422060256e-06,
|
| 3053 |
+
"loss": 0.4416,
|
| 3054 |
+
"step": 435
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 4.607661822985469,
|
| 3058 |
+
"grad_norm": 0.0920008670611069,
|
| 3059 |
+
"learning_rate": 1.2685217263972693e-06,
|
| 3060 |
+
"loss": 0.4433,
|
| 3061 |
+
"step": 436
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 4.618229854689564,
|
| 3065 |
+
"grad_norm": 0.08954221258416922,
|
| 3066 |
+
"learning_rate": 1.1953686032990964e-06,
|
| 3067 |
+
"loss": 0.4431,
|
| 3068 |
+
"step": 437
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 4.628797886393659,
|
| 3072 |
+
"grad_norm": 0.0897729908602636,
|
| 3073 |
+
"learning_rate": 1.124355907970487e-06,
|
| 3074 |
+
"loss": 0.4451,
|
| 3075 |
+
"step": 438
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 4.639365918097754,
|
| 3079 |
+
"grad_norm": 0.09192361815684204,
|
| 3080 |
+
"learning_rate": 1.0554875574063784e-06,
|
| 3081 |
+
"loss": 0.4423,
|
| 3082 |
+
"step": 439
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 4.6499339498018495,
|
| 3086 |
+
"grad_norm": 0.09033170522999512,
|
| 3087 |
+
"learning_rate": 9.887673503216422e-07,
|
| 3088 |
+
"loss": 0.4394,
|
| 3089 |
+
"step": 440
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 4.660501981505945,
|
| 3093 |
+
"grad_norm": 0.09133489415949655,
|
| 3094 |
+
"learning_rate": 9.241989669415097e-07,
|
| 3095 |
+
"loss": 0.4447,
|
| 3096 |
+
"step": 441
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 4.67107001321004,
|
| 3100 |
+
"grad_norm": 0.09425014878930163,
|
| 3101 |
+
"learning_rate": 8.61785968798623e-07,
|
| 3102 |
+
"loss": 0.4397,
|
| 3103 |
+
"step": 442
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 4.681638044914135,
|
| 3107 |
+
"grad_norm": 0.09161607478315605,
|
| 3108 |
+
"learning_rate": 8.015317985365301e-07,
|
| 3109 |
+
"loss": 0.4435,
|
| 3110 |
+
"step": 443
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 4.69220607661823,
|
| 3114 |
+
"grad_norm": 0.0915131016919999,
|
| 3115 |
+
"learning_rate": 7.434397797198367e-07,
|
| 3116 |
+
"loss": 0.4392,
|
| 3117 |
+
"step": 444
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 4.702774108322325,
|
| 3121 |
+
"grad_norm": 0.09011982457022855,
|
| 3122 |
+
"learning_rate": 6.875131166508553e-07,
|
| 3123 |
+
"loss": 0.4429,
|
| 3124 |
+
"step": 445
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 4.71334214002642,
|
| 3128 |
+
"grad_norm": 0.09105298330396881,
|
| 3129 |
+
"learning_rate": 6.337548941928839e-07,
|
| 3130 |
+
"loss": 0.4452,
|
| 3131 |
+
"step": 446
|
| 3132 |
+
},
|
| 3133 |
+
{
|
| 3134 |
+
"epoch": 4.723910171730515,
|
| 3135 |
+
"grad_norm": 0.08866673271037578,
|
| 3136 |
+
"learning_rate": 5.821680776000049e-07,
|
| 3137 |
+
"loss": 0.4422,
|
| 3138 |
+
"step": 447
|
| 3139 |
+
},
|
| 3140 |
+
{
|
| 3141 |
+
"epoch": 4.734478203434611,
|
| 3142 |
+
"grad_norm": 0.08801181304315894,
|
| 3143 |
+
"learning_rate": 5.32755512353571e-07,
|
| 3144 |
+
"loss": 0.436,
|
| 3145 |
+
"step": 448
|
| 3146 |
+
},
|
| 3147 |
+
{
|
| 3148 |
+
"epoch": 4.745046235138705,
|
| 3149 |
+
"grad_norm": 0.0892771815810736,
|
| 3150 |
+
"learning_rate": 4.8551992400522e-07,
|
| 3151 |
+
"loss": 0.4452,
|
| 3152 |
+
"step": 449
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 4.755614266842801,
|
| 3156 |
+
"grad_norm": 0.08829094089099762,
|
| 3157 |
+
"learning_rate": 4.4046391802655463e-07,
|
| 3158 |
+
"loss": 0.4422,
|
| 3159 |
+
"step": 450
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 4.766182298546895,
|
| 3163 |
+
"grad_norm": 0.09129909001389838,
|
| 3164 |
+
"learning_rate": 3.975899796654137e-07,
|
| 3165 |
+
"loss": 0.443,
|
| 3166 |
+
"step": 451
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 4.776750330250991,
|
| 3170 |
+
"grad_norm": 0.0898229105447953,
|
| 3171 |
+
"learning_rate": 3.569004738087989e-07,
|
| 3172 |
+
"loss": 0.4408,
|
| 3173 |
+
"step": 452
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 4.787318361955085,
|
| 3177 |
+
"grad_norm": 0.08880239621448514,
|
| 3178 |
+
"learning_rate": 3.183976448524106e-07,
|
| 3179 |
+
"loss": 0.4375,
|
| 3180 |
+
"step": 453
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 4.797886393659181,
|
| 3184 |
+
"grad_norm": 0.08857557909002102,
|
| 3185 |
+
"learning_rate": 2.8208361657688474e-07,
|
| 3186 |
+
"loss": 0.4422,
|
| 3187 |
+
"step": 454
|
| 3188 |
+
},
|
| 3189 |
+
{
|
| 3190 |
+
"epoch": 4.808454425363276,
|
| 3191 |
+
"grad_norm": 0.08766599065540119,
|
| 3192 |
+
"learning_rate": 2.479603920306106e-07,
|
| 3193 |
+
"loss": 0.4358,
|
| 3194 |
+
"step": 455
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 4.819022457067371,
|
| 3198 |
+
"grad_norm": 0.09087441316527663,
|
| 3199 |
+
"learning_rate": 2.1602985341925953e-07,
|
| 3200 |
+
"loss": 0.4366,
|
| 3201 |
+
"step": 456
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 4.8295904887714665,
|
| 3205 |
+
"grad_norm": 0.0907938224300551,
|
| 3206 |
+
"learning_rate": 1.8629376200197004e-07,
|
| 3207 |
+
"loss": 0.442,
|
| 3208 |
+
"step": 457
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 4.840158520475561,
|
| 3212 |
+
"grad_norm": 0.08888968582423885,
|
| 3213 |
+
"learning_rate": 1.5875375799419e-07,
|
| 3214 |
+
"loss": 0.4358,
|
| 3215 |
+
"step": 458
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 4.8507265521796565,
|
| 3219 |
+
"grad_norm": 0.08900710504460911,
|
| 3220 |
+
"learning_rate": 1.3341136047719805e-07,
|
| 3221 |
+
"loss": 0.4367,
|
| 3222 |
+
"step": 459
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 4.861294583883752,
|
| 3226 |
+
"grad_norm": 0.08986360536362852,
|
| 3227 |
+
"learning_rate": 1.1026796731433475e-07,
|
| 3228 |
+
"loss": 0.4414,
|
| 3229 |
+
"step": 460
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 4.871862615587847,
|
| 3233 |
+
"grad_norm": 0.08885451973141226,
|
| 3234 |
+
"learning_rate": 8.932485507387345e-08,
|
| 3235 |
+
"loss": 0.4368,
|
| 3236 |
+
"step": 461
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 4.882430647291942,
|
| 3240 |
+
"grad_norm": 0.08728339714966311,
|
| 3241 |
+
"learning_rate": 7.058317895861866e-08,
|
| 3242 |
+
"loss": 0.445,
|
| 3243 |
+
"step": 462
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 4.892998678996037,
|
| 3247 |
+
"grad_norm": 0.08758661927375506,
|
| 3248 |
+
"learning_rate": 5.40439727421882e-08,
|
| 3249 |
+
"loss": 0.442,
|
| 3250 |
+
"step": 463
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 4.903566710700132,
|
| 3254 |
+
"grad_norm": 0.0906444113153008,
|
| 3255 |
+
"learning_rate": 3.970814871197437e-08,
|
| 3256 |
+
"loss": 0.4387,
|
| 3257 |
+
"step": 464
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 4.914134742404228,
|
| 3261 |
+
"grad_norm": 0.08754952023250234,
|
| 3262 |
+
"learning_rate": 2.75764976188464e-08,
|
| 3263 |
+
"loss": 0.4401,
|
| 3264 |
+
"step": 465
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 4.924702774108322,
|
| 3268 |
+
"grad_norm": 0.08803257540350204,
|
| 3269 |
+
"learning_rate": 1.764968863351424e-08,
|
| 3270 |
+
"loss": 0.4401,
|
| 3271 |
+
"step": 466
|
| 3272 |
+
},
|
| 3273 |
+
{
|
| 3274 |
+
"epoch": 4.935270805812418,
|
| 3275 |
+
"grad_norm": 0.08863746420839871,
|
| 3276 |
+
"learning_rate": 9.928269309638083e-09,
|
| 3277 |
+
"loss": 0.4398,
|
| 3278 |
+
"step": 467
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 4.945838837516512,
|
| 3282 |
+
"grad_norm": 0.08825690980969667,
|
| 3283 |
+
"learning_rate": 4.412665553594764e-09,
|
| 3284 |
+
"loss": 0.4478,
|
| 3285 |
+
"step": 468
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 4.956406869220608,
|
| 3289 |
+
"grad_norm": 0.09184591938217317,
|
| 3290 |
+
"learning_rate": 1.103181601020964e-09,
|
| 3291 |
+
"loss": 0.4439,
|
| 3292 |
+
"step": 469
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 4.966974900924702,
|
| 3296 |
+
"grad_norm": 0.08777993584647581,
|
| 3297 |
+
"learning_rate": 0.0,
|
| 3298 |
+
"loss": 0.4386,
|
| 3299 |
+
"step": 470
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 4.966974900924702,
|
| 3303 |
+
"step": 470,
|
| 3304 |
+
"total_flos": 1.2094498954511647e+19,
|
| 3305 |
+
"train_loss": 0.5709579251548077,
|
| 3306 |
+
"train_runtime": 47190.5732,
|
| 3307 |
+
"train_samples_per_second": 5.131,
|
| 3308 |
+
"train_steps_per_second": 0.01
|
| 3309 |
+
}
|
| 3310 |
+
],
|
| 3311 |
+
"logging_steps": 1,
|
| 3312 |
+
"max_steps": 470,
|
| 3313 |
+
"num_input_tokens_seen": 0,
|
| 3314 |
+
"num_train_epochs": 5,
|
| 3315 |
+
"save_steps": 500,
|
| 3316 |
+
"stateful_callbacks": {
|
| 3317 |
+
"TrainerControl": {
|
| 3318 |
+
"args": {
|
| 3319 |
+
"should_epoch_stop": false,
|
| 3320 |
+
"should_evaluate": false,
|
| 3321 |
+
"should_log": false,
|
| 3322 |
+
"should_save": true,
|
| 3323 |
+
"should_training_stop": true
|
| 3324 |
+
},
|
| 3325 |
+
"attributes": {}
|
| 3326 |
+
}
|
| 3327 |
+
},
|
| 3328 |
+
"total_flos": 1.2094498954511647e+19,
|
| 3329 |
+
"train_batch_size": 1,
|
| 3330 |
+
"trial_name": null,
|
| 3331 |
+
"trial_params": null
|
| 3332 |
+
}
|
training_loss.png
ADDED
|