End of training
Browse files- README.md +2 -1
- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +3696 -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: global_batchsize_96_lr2e5
|
|
@@ -15,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# global_batchsize_96_lr2e5
|
| 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: global_batchsize_96_lr2e5
|
|
|
|
| 16 |
|
| 17 |
# global_batchsize_96_lr2e5
|
| 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 bespokelabs/Bespoke-Stratos-17k dataset.
|
| 20 |
|
| 21 |
## Model description
|
| 22 |
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.994263862332696,
|
| 3 |
+
"total_flos": 5.620362022127534e+17,
|
| 4 |
+
"train_loss": 0.45750690551324824,
|
| 5 |
+
"train_runtime": 8551.5314,
|
| 6 |
+
"train_samples_per_second": 5.862,
|
| 7 |
+
"train_steps_per_second": 0.061
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.994263862332696,
|
| 3 |
+
"total_flos": 5.620362022127534e+17,
|
| 4 |
+
"train_loss": 0.45750690551324824,
|
| 5 |
+
"train_runtime": 8551.5314,
|
| 6 |
+
"train_samples_per_second": 5.862,
|
| 7 |
+
"train_steps_per_second": 0.061
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3696 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.994263862332696,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 522,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0057361376673040155,
|
| 13 |
+
"grad_norm": 5.963601607855415,
|
| 14 |
+
"learning_rate": 3.773584905660378e-07,
|
| 15 |
+
"loss": 0.8872,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.011472275334608031,
|
| 20 |
+
"grad_norm": 6.333703884001195,
|
| 21 |
+
"learning_rate": 7.547169811320755e-07,
|
| 22 |
+
"loss": 0.9165,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.017208413001912046,
|
| 27 |
+
"grad_norm": 6.257855070229476,
|
| 28 |
+
"learning_rate": 1.1320754716981133e-06,
|
| 29 |
+
"loss": 0.9418,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.022944550669216062,
|
| 34 |
+
"grad_norm": 5.956524172527457,
|
| 35 |
+
"learning_rate": 1.509433962264151e-06,
|
| 36 |
+
"loss": 0.9044,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.028680688336520075,
|
| 41 |
+
"grad_norm": 5.705771711698959,
|
| 42 |
+
"learning_rate": 1.8867924528301889e-06,
|
| 43 |
+
"loss": 0.8938,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.03441682600382409,
|
| 48 |
+
"grad_norm": 5.8492677879174275,
|
| 49 |
+
"learning_rate": 2.2641509433962266e-06,
|
| 50 |
+
"loss": 0.8815,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.040152963671128104,
|
| 55 |
+
"grad_norm": 4.727616456076574,
|
| 56 |
+
"learning_rate": 2.6415094339622644e-06,
|
| 57 |
+
"loss": 0.8718,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.045889101338432124,
|
| 62 |
+
"grad_norm": 2.6576351181409055,
|
| 63 |
+
"learning_rate": 3.018867924528302e-06,
|
| 64 |
+
"loss": 0.8163,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.05162523900573614,
|
| 69 |
+
"grad_norm": 2.444013780371823,
|
| 70 |
+
"learning_rate": 3.3962264150943395e-06,
|
| 71 |
+
"loss": 0.7805,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.05736137667304015,
|
| 76 |
+
"grad_norm": 2.190871953398052,
|
| 77 |
+
"learning_rate": 3.7735849056603777e-06,
|
| 78 |
+
"loss": 0.8185,
|
| 79 |
+
"step": 10
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.06309751434034416,
|
| 83 |
+
"grad_norm": 3.5229712196521215,
|
| 84 |
+
"learning_rate": 4.150943396226416e-06,
|
| 85 |
+
"loss": 0.7837,
|
| 86 |
+
"step": 11
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.06883365200764818,
|
| 90 |
+
"grad_norm": 3.726026713435741,
|
| 91 |
+
"learning_rate": 4.528301886792453e-06,
|
| 92 |
+
"loss": 0.733,
|
| 93 |
+
"step": 12
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.0745697896749522,
|
| 97 |
+
"grad_norm": 4.108103957749213,
|
| 98 |
+
"learning_rate": 4.905660377358491e-06,
|
| 99 |
+
"loss": 0.7995,
|
| 100 |
+
"step": 13
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.08030592734225621,
|
| 104 |
+
"grad_norm": 3.669645529053506,
|
| 105 |
+
"learning_rate": 5.283018867924529e-06,
|
| 106 |
+
"loss": 0.7584,
|
| 107 |
+
"step": 14
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.08604206500956023,
|
| 111 |
+
"grad_norm": 2.952547188045258,
|
| 112 |
+
"learning_rate": 5.660377358490566e-06,
|
| 113 |
+
"loss": 0.7447,
|
| 114 |
+
"step": 15
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.09177820267686425,
|
| 118 |
+
"grad_norm": 2.398455043486125,
|
| 119 |
+
"learning_rate": 6.037735849056604e-06,
|
| 120 |
+
"loss": 0.74,
|
| 121 |
+
"step": 16
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.09751434034416825,
|
| 125 |
+
"grad_norm": 1.829688656727974,
|
| 126 |
+
"learning_rate": 6.415094339622642e-06,
|
| 127 |
+
"loss": 0.7052,
|
| 128 |
+
"step": 17
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.10325047801147227,
|
| 132 |
+
"grad_norm": 1.7012819538641355,
|
| 133 |
+
"learning_rate": 6.792452830188679e-06,
|
| 134 |
+
"loss": 0.7006,
|
| 135 |
+
"step": 18
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.1089866156787763,
|
| 139 |
+
"grad_norm": 1.5287783620949145,
|
| 140 |
+
"learning_rate": 7.169811320754717e-06,
|
| 141 |
+
"loss": 0.6299,
|
| 142 |
+
"step": 19
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.1147227533460803,
|
| 146 |
+
"grad_norm": 1.6890797388961225,
|
| 147 |
+
"learning_rate": 7.5471698113207555e-06,
|
| 148 |
+
"loss": 0.6694,
|
| 149 |
+
"step": 20
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.12045889101338432,
|
| 153 |
+
"grad_norm": 1.5848500020422387,
|
| 154 |
+
"learning_rate": 7.924528301886793e-06,
|
| 155 |
+
"loss": 0.6481,
|
| 156 |
+
"step": 21
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.12619502868068833,
|
| 160 |
+
"grad_norm": 1.2651922200399228,
|
| 161 |
+
"learning_rate": 8.301886792452832e-06,
|
| 162 |
+
"loss": 0.643,
|
| 163 |
+
"step": 22
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.13193116634799235,
|
| 167 |
+
"grad_norm": 1.1704521558620569,
|
| 168 |
+
"learning_rate": 8.67924528301887e-06,
|
| 169 |
+
"loss": 0.6076,
|
| 170 |
+
"step": 23
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.13766730401529637,
|
| 174 |
+
"grad_norm": 1.211023371219137,
|
| 175 |
+
"learning_rate": 9.056603773584907e-06,
|
| 176 |
+
"loss": 0.6383,
|
| 177 |
+
"step": 24
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.14340344168260039,
|
| 181 |
+
"grad_norm": 1.0419000696186218,
|
| 182 |
+
"learning_rate": 9.433962264150944e-06,
|
| 183 |
+
"loss": 0.6244,
|
| 184 |
+
"step": 25
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.1491395793499044,
|
| 188 |
+
"grad_norm": 1.0173864861869972,
|
| 189 |
+
"learning_rate": 9.811320754716981e-06,
|
| 190 |
+
"loss": 0.6012,
|
| 191 |
+
"step": 26
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.15487571701720843,
|
| 195 |
+
"grad_norm": 1.1096264531149542,
|
| 196 |
+
"learning_rate": 1.018867924528302e-05,
|
| 197 |
+
"loss": 0.6421,
|
| 198 |
+
"step": 27
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.16061185468451242,
|
| 202 |
+
"grad_norm": 0.9517151993809669,
|
| 203 |
+
"learning_rate": 1.0566037735849058e-05,
|
| 204 |
+
"loss": 0.6205,
|
| 205 |
+
"step": 28
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.16634799235181644,
|
| 209 |
+
"grad_norm": 0.7532883579581415,
|
| 210 |
+
"learning_rate": 1.0943396226415095e-05,
|
| 211 |
+
"loss": 0.6032,
|
| 212 |
+
"step": 29
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.17208413001912046,
|
| 216 |
+
"grad_norm": 0.8143750313574308,
|
| 217 |
+
"learning_rate": 1.1320754716981132e-05,
|
| 218 |
+
"loss": 0.6013,
|
| 219 |
+
"step": 30
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.17782026768642448,
|
| 223 |
+
"grad_norm": 0.8204477456168644,
|
| 224 |
+
"learning_rate": 1.169811320754717e-05,
|
| 225 |
+
"loss": 0.5721,
|
| 226 |
+
"step": 31
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.1835564053537285,
|
| 230 |
+
"grad_norm": 0.7275017869334657,
|
| 231 |
+
"learning_rate": 1.2075471698113209e-05,
|
| 232 |
+
"loss": 0.5571,
|
| 233 |
+
"step": 32
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.18929254302103252,
|
| 237 |
+
"grad_norm": 0.6610654194161306,
|
| 238 |
+
"learning_rate": 1.2452830188679246e-05,
|
| 239 |
+
"loss": 0.5833,
|
| 240 |
+
"step": 33
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.1950286806883365,
|
| 244 |
+
"grad_norm": 0.6495136985020462,
|
| 245 |
+
"learning_rate": 1.2830188679245283e-05,
|
| 246 |
+
"loss": 0.6104,
|
| 247 |
+
"step": 34
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.20076481835564053,
|
| 251 |
+
"grad_norm": 0.6949029178393563,
|
| 252 |
+
"learning_rate": 1.320754716981132e-05,
|
| 253 |
+
"loss": 0.5837,
|
| 254 |
+
"step": 35
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.20650095602294455,
|
| 258 |
+
"grad_norm": 0.6489044947838399,
|
| 259 |
+
"learning_rate": 1.3584905660377358e-05,
|
| 260 |
+
"loss": 0.5722,
|
| 261 |
+
"step": 36
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.21223709369024857,
|
| 265 |
+
"grad_norm": 0.5408166196721413,
|
| 266 |
+
"learning_rate": 1.3962264150943397e-05,
|
| 267 |
+
"loss": 0.5711,
|
| 268 |
+
"step": 37
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.2179732313575526,
|
| 272 |
+
"grad_norm": 0.7281447634302773,
|
| 273 |
+
"learning_rate": 1.4339622641509435e-05,
|
| 274 |
+
"loss": 0.5892,
|
| 275 |
+
"step": 38
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.2237093690248566,
|
| 279 |
+
"grad_norm": 0.6166840085962729,
|
| 280 |
+
"learning_rate": 1.4716981132075472e-05,
|
| 281 |
+
"loss": 0.5924,
|
| 282 |
+
"step": 39
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.2294455066921606,
|
| 286 |
+
"grad_norm": 0.6040727735751141,
|
| 287 |
+
"learning_rate": 1.5094339622641511e-05,
|
| 288 |
+
"loss": 0.5446,
|
| 289 |
+
"step": 40
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.23518164435946462,
|
| 293 |
+
"grad_norm": 0.6273113840367965,
|
| 294 |
+
"learning_rate": 1.547169811320755e-05,
|
| 295 |
+
"loss": 0.5982,
|
| 296 |
+
"step": 41
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.24091778202676864,
|
| 300 |
+
"grad_norm": 0.6363659081534939,
|
| 301 |
+
"learning_rate": 1.5849056603773586e-05,
|
| 302 |
+
"loss": 0.5336,
|
| 303 |
+
"step": 42
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.24665391969407266,
|
| 307 |
+
"grad_norm": 0.6615133633303231,
|
| 308 |
+
"learning_rate": 1.6226415094339625e-05,
|
| 309 |
+
"loss": 0.5879,
|
| 310 |
+
"step": 43
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.25239005736137665,
|
| 314 |
+
"grad_norm": 0.8414155664286366,
|
| 315 |
+
"learning_rate": 1.6603773584905664e-05,
|
| 316 |
+
"loss": 0.5854,
|
| 317 |
+
"step": 44
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.25812619502868067,
|
| 321 |
+
"grad_norm": 0.6770822965707434,
|
| 322 |
+
"learning_rate": 1.69811320754717e-05,
|
| 323 |
+
"loss": 0.5684,
|
| 324 |
+
"step": 45
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.2638623326959847,
|
| 328 |
+
"grad_norm": 0.562245948225568,
|
| 329 |
+
"learning_rate": 1.735849056603774e-05,
|
| 330 |
+
"loss": 0.5275,
|
| 331 |
+
"step": 46
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.2695984703632887,
|
| 335 |
+
"grad_norm": 0.7981141857389045,
|
| 336 |
+
"learning_rate": 1.7735849056603774e-05,
|
| 337 |
+
"loss": 0.5689,
|
| 338 |
+
"step": 47
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.27533460803059273,
|
| 342 |
+
"grad_norm": 0.5837129815497581,
|
| 343 |
+
"learning_rate": 1.8113207547169813e-05,
|
| 344 |
+
"loss": 0.5544,
|
| 345 |
+
"step": 48
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.28107074569789675,
|
| 349 |
+
"grad_norm": 0.6069344726727888,
|
| 350 |
+
"learning_rate": 1.8490566037735852e-05,
|
| 351 |
+
"loss": 0.5788,
|
| 352 |
+
"step": 49
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.28680688336520077,
|
| 356 |
+
"grad_norm": 0.5964572258571391,
|
| 357 |
+
"learning_rate": 1.8867924528301888e-05,
|
| 358 |
+
"loss": 0.5352,
|
| 359 |
+
"step": 50
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.2925430210325048,
|
| 363 |
+
"grad_norm": 4.690516652663125,
|
| 364 |
+
"learning_rate": 1.9245283018867927e-05,
|
| 365 |
+
"loss": 0.5248,
|
| 366 |
+
"step": 51
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.2982791586998088,
|
| 370 |
+
"grad_norm": 0.8872735078180753,
|
| 371 |
+
"learning_rate": 1.9622641509433963e-05,
|
| 372 |
+
"loss": 0.5665,
|
| 373 |
+
"step": 52
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.30401529636711283,
|
| 377 |
+
"grad_norm": 0.5723834462499225,
|
| 378 |
+
"learning_rate": 2e-05,
|
| 379 |
+
"loss": 0.5539,
|
| 380 |
+
"step": 53
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.30975143403441685,
|
| 384 |
+
"grad_norm": 0.6524504508920996,
|
| 385 |
+
"learning_rate": 1.999977565187699e-05,
|
| 386 |
+
"loss": 0.5232,
|
| 387 |
+
"step": 54
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.3154875717017208,
|
| 391 |
+
"grad_norm": 0.6468217495098721,
|
| 392 |
+
"learning_rate": 1.9999102617574366e-05,
|
| 393 |
+
"loss": 0.5387,
|
| 394 |
+
"step": 55
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.32122370936902483,
|
| 398 |
+
"grad_norm": 0.6744222328381954,
|
| 399 |
+
"learning_rate": 1.9997980927290928e-05,
|
| 400 |
+
"loss": 0.5461,
|
| 401 |
+
"step": 56
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.32695984703632885,
|
| 405 |
+
"grad_norm": 0.7945647096558187,
|
| 406 |
+
"learning_rate": 1.9996410631356496e-05,
|
| 407 |
+
"loss": 0.5501,
|
| 408 |
+
"step": 57
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.3326959847036329,
|
| 412 |
+
"grad_norm": 0.583537979504518,
|
| 413 |
+
"learning_rate": 1.9994391800229666e-05,
|
| 414 |
+
"loss": 0.5475,
|
| 415 |
+
"step": 58
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.3384321223709369,
|
| 419 |
+
"grad_norm": 0.7688713046343357,
|
| 420 |
+
"learning_rate": 1.999192452449463e-05,
|
| 421 |
+
"loss": 0.5341,
|
| 422 |
+
"step": 59
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.3441682600382409,
|
| 426 |
+
"grad_norm": 0.6225461236118923,
|
| 427 |
+
"learning_rate": 1.9989008914857115e-05,
|
| 428 |
+
"loss": 0.5215,
|
| 429 |
+
"step": 60
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.34990439770554493,
|
| 433 |
+
"grad_norm": 0.7556986010825483,
|
| 434 |
+
"learning_rate": 1.998564510213944e-05,
|
| 435 |
+
"loss": 0.5485,
|
| 436 |
+
"step": 61
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.35564053537284895,
|
| 440 |
+
"grad_norm": 0.6466498188523241,
|
| 441 |
+
"learning_rate": 1.998183323727462e-05,
|
| 442 |
+
"loss": 0.5472,
|
| 443 |
+
"step": 62
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.361376673040153,
|
| 447 |
+
"grad_norm": 0.6498839386655383,
|
| 448 |
+
"learning_rate": 1.9977573491299597e-05,
|
| 449 |
+
"loss": 0.5458,
|
| 450 |
+
"step": 63
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.367112810707457,
|
| 454 |
+
"grad_norm": 0.7893514486845328,
|
| 455 |
+
"learning_rate": 1.9972866055347572e-05,
|
| 456 |
+
"loss": 0.5268,
|
| 457 |
+
"step": 64
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.372848948374761,
|
| 461 |
+
"grad_norm": 0.7956885248224083,
|
| 462 |
+
"learning_rate": 1.996771114063943e-05,
|
| 463 |
+
"loss": 0.5388,
|
| 464 |
+
"step": 65
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.37858508604206503,
|
| 468 |
+
"grad_norm": 0.8314332190289494,
|
| 469 |
+
"learning_rate": 1.9962108978474265e-05,
|
| 470 |
+
"loss": 0.5431,
|
| 471 |
+
"step": 66
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.384321223709369,
|
| 475 |
+
"grad_norm": 0.8426823703479597,
|
| 476 |
+
"learning_rate": 1.9956059820218982e-05,
|
| 477 |
+
"loss": 0.5151,
|
| 478 |
+
"step": 67
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.390057361376673,
|
| 482 |
+
"grad_norm": 0.7757061162171306,
|
| 483 |
+
"learning_rate": 1.9949563937297045e-05,
|
| 484 |
+
"loss": 0.5268,
|
| 485 |
+
"step": 68
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.39579349904397704,
|
| 489 |
+
"grad_norm": 2.4653347776714294,
|
| 490 |
+
"learning_rate": 1.9942621621176283e-05,
|
| 491 |
+
"loss": 0.5349,
|
| 492 |
+
"step": 69
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.40152963671128106,
|
| 496 |
+
"grad_norm": 0.8709833395015452,
|
| 497 |
+
"learning_rate": 1.993523318335581e-05,
|
| 498 |
+
"loss": 0.4825,
|
| 499 |
+
"step": 70
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.4072657743785851,
|
| 503 |
+
"grad_norm": 9.961649749835258,
|
| 504 |
+
"learning_rate": 1.9927398955352062e-05,
|
| 505 |
+
"loss": 0.5363,
|
| 506 |
+
"step": 71
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.4130019120458891,
|
| 510 |
+
"grad_norm": 0.9561541581658524,
|
| 511 |
+
"learning_rate": 1.9919119288683908e-05,
|
| 512 |
+
"loss": 0.504,
|
| 513 |
+
"step": 72
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.4187380497131931,
|
| 517 |
+
"grad_norm": 0.680622575434205,
|
| 518 |
+
"learning_rate": 1.991039455485688e-05,
|
| 519 |
+
"loss": 0.5364,
|
| 520 |
+
"step": 73
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.42447418738049714,
|
| 524 |
+
"grad_norm": 0.8693452396686029,
|
| 525 |
+
"learning_rate": 1.990122514534651e-05,
|
| 526 |
+
"loss": 0.5331,
|
| 527 |
+
"step": 74
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.43021032504780116,
|
| 531 |
+
"grad_norm": 0.7743910198858713,
|
| 532 |
+
"learning_rate": 1.9891611471580767e-05,
|
| 533 |
+
"loss": 0.5496,
|
| 534 |
+
"step": 75
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.4359464627151052,
|
| 538 |
+
"grad_norm": 0.8765835472970227,
|
| 539 |
+
"learning_rate": 1.9881553964921574e-05,
|
| 540 |
+
"loss": 0.5313,
|
| 541 |
+
"step": 76
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.4416826003824092,
|
| 545 |
+
"grad_norm": 0.7689840181875675,
|
| 546 |
+
"learning_rate": 1.987105307664549e-05,
|
| 547 |
+
"loss": 0.544,
|
| 548 |
+
"step": 77
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.4474187380497132,
|
| 552 |
+
"grad_norm": 0.6520437505199477,
|
| 553 |
+
"learning_rate": 1.9860109277923417e-05,
|
| 554 |
+
"loss": 0.5314,
|
| 555 |
+
"step": 78
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.45315487571701724,
|
| 559 |
+
"grad_norm": 0.8673404414655712,
|
| 560 |
+
"learning_rate": 1.9848723059799508e-05,
|
| 561 |
+
"loss": 0.5429,
|
| 562 |
+
"step": 79
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.4588910133843212,
|
| 566 |
+
"grad_norm": 0.6109886586522215,
|
| 567 |
+
"learning_rate": 1.983689493316909e-05,
|
| 568 |
+
"loss": 0.5223,
|
| 569 |
+
"step": 80
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.4646271510516252,
|
| 573 |
+
"grad_norm": 0.7103480782598592,
|
| 574 |
+
"learning_rate": 1.982462542875576e-05,
|
| 575 |
+
"loss": 0.5277,
|
| 576 |
+
"step": 81
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.47036328871892924,
|
| 580 |
+
"grad_norm": 0.6705871728122585,
|
| 581 |
+
"learning_rate": 1.9811915097087587e-05,
|
| 582 |
+
"loss": 0.5496,
|
| 583 |
+
"step": 82
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.47609942638623326,
|
| 587 |
+
"grad_norm": 0.7785921338601758,
|
| 588 |
+
"learning_rate": 1.9798764508472373e-05,
|
| 589 |
+
"loss": 0.5234,
|
| 590 |
+
"step": 83
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.4818355640535373,
|
| 594 |
+
"grad_norm": 0.7439955422623991,
|
| 595 |
+
"learning_rate": 1.9785174252972092e-05,
|
| 596 |
+
"loss": 0.5361,
|
| 597 |
+
"step": 84
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.4875717017208413,
|
| 601 |
+
"grad_norm": 0.797568241880013,
|
| 602 |
+
"learning_rate": 1.9771144940376413e-05,
|
| 603 |
+
"loss": 0.5214,
|
| 604 |
+
"step": 85
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.4933078393881453,
|
| 608 |
+
"grad_norm": 0.6230345386937433,
|
| 609 |
+
"learning_rate": 1.9756677200175316e-05,
|
| 610 |
+
"loss": 0.5261,
|
| 611 |
+
"step": 86
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.49904397705544934,
|
| 615 |
+
"grad_norm": 0.7261095628950723,
|
| 616 |
+
"learning_rate": 1.974177168153088e-05,
|
| 617 |
+
"loss": 0.5485,
|
| 618 |
+
"step": 87
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.5047801147227533,
|
| 622 |
+
"grad_norm": 0.5411293935713085,
|
| 623 |
+
"learning_rate": 1.972642905324813e-05,
|
| 624 |
+
"loss": 0.5281,
|
| 625 |
+
"step": 88
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.5105162523900574,
|
| 629 |
+
"grad_norm": 0.6298816468045366,
|
| 630 |
+
"learning_rate": 1.971065000374504e-05,
|
| 631 |
+
"loss": 0.5372,
|
| 632 |
+
"step": 89
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.5162523900573613,
|
| 636 |
+
"grad_norm": 0.4908410042540129,
|
| 637 |
+
"learning_rate": 1.969443524102163e-05,
|
| 638 |
+
"loss": 0.5429,
|
| 639 |
+
"step": 90
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.5219885277246654,
|
| 643 |
+
"grad_norm": 0.6557029434687354,
|
| 644 |
+
"learning_rate": 1.967778549262822e-05,
|
| 645 |
+
"loss": 0.497,
|
| 646 |
+
"step": 91
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.5277246653919694,
|
| 650 |
+
"grad_norm": 0.5620880029295425,
|
| 651 |
+
"learning_rate": 1.9660701505632773e-05,
|
| 652 |
+
"loss": 0.514,
|
| 653 |
+
"step": 92
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.5334608030592735,
|
| 657 |
+
"grad_norm": 0.6884951285963853,
|
| 658 |
+
"learning_rate": 1.9643184046587373e-05,
|
| 659 |
+
"loss": 0.5239,
|
| 660 |
+
"step": 93
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.5391969407265774,
|
| 664 |
+
"grad_norm": 0.6604434604547247,
|
| 665 |
+
"learning_rate": 1.962523390149382e-05,
|
| 666 |
+
"loss": 0.5207,
|
| 667 |
+
"step": 94
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.5449330783938815,
|
| 671 |
+
"grad_norm": 0.7336652529710372,
|
| 672 |
+
"learning_rate": 1.9606851875768404e-05,
|
| 673 |
+
"loss": 0.5331,
|
| 674 |
+
"step": 95
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.5506692160611855,
|
| 678 |
+
"grad_norm": 0.6669732222358924,
|
| 679 |
+
"learning_rate": 1.9588038794205705e-05,
|
| 680 |
+
"loss": 0.5323,
|
| 681 |
+
"step": 96
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.5564053537284895,
|
| 685 |
+
"grad_norm": 0.6137757201982459,
|
| 686 |
+
"learning_rate": 1.9568795500941635e-05,
|
| 687 |
+
"loss": 0.5023,
|
| 688 |
+
"step": 97
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.5621414913957935,
|
| 692 |
+
"grad_norm": 0.6272322569320127,
|
| 693 |
+
"learning_rate": 1.9549122859415538e-05,
|
| 694 |
+
"loss": 0.5576,
|
| 695 |
+
"step": 98
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.5678776290630975,
|
| 699 |
+
"grad_norm": 0.5620575180386782,
|
| 700 |
+
"learning_rate": 1.9529021752331455e-05,
|
| 701 |
+
"loss": 0.5186,
|
| 702 |
+
"step": 99
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.5736137667304015,
|
| 706 |
+
"grad_norm": 0.5651887008267796,
|
| 707 |
+
"learning_rate": 1.9508493081618515e-05,
|
| 708 |
+
"loss": 0.5105,
|
| 709 |
+
"step": 100
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.5793499043977055,
|
| 713 |
+
"grad_norm": 0.5618076420744421,
|
| 714 |
+
"learning_rate": 1.9487537768390465e-05,
|
| 715 |
+
"loss": 0.5406,
|
| 716 |
+
"step": 101
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.5850860420650096,
|
| 720 |
+
"grad_norm": 0.7480174415686112,
|
| 721 |
+
"learning_rate": 1.9466156752904344e-05,
|
| 722 |
+
"loss": 0.5351,
|
| 723 |
+
"step": 102
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.5908221797323135,
|
| 727 |
+
"grad_norm": 0.507106651617815,
|
| 728 |
+
"learning_rate": 1.944435099451829e-05,
|
| 729 |
+
"loss": 0.5266,
|
| 730 |
+
"step": 103
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.5965583173996176,
|
| 734 |
+
"grad_norm": 0.7081327554157436,
|
| 735 |
+
"learning_rate": 1.94221214716485e-05,
|
| 736 |
+
"loss": 0.5181,
|
| 737 |
+
"step": 104
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.6022944550669216,
|
| 741 |
+
"grad_norm": 0.5232017157016485,
|
| 742 |
+
"learning_rate": 1.939946918172531e-05,
|
| 743 |
+
"loss": 0.5114,
|
| 744 |
+
"step": 105
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.6080305927342257,
|
| 748 |
+
"grad_norm": 0.5400711534025171,
|
| 749 |
+
"learning_rate": 1.9376395141148475e-05,
|
| 750 |
+
"loss": 0.538,
|
| 751 |
+
"step": 106
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.6137667304015296,
|
| 755 |
+
"grad_norm": 0.5469717806295151,
|
| 756 |
+
"learning_rate": 1.9352900385241534e-05,
|
| 757 |
+
"loss": 0.5237,
|
| 758 |
+
"step": 107
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.6195028680688337,
|
| 762 |
+
"grad_norm": 0.44841617252688876,
|
| 763 |
+
"learning_rate": 1.932898596820536e-05,
|
| 764 |
+
"loss": 0.5315,
|
| 765 |
+
"step": 108
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.6252390057361377,
|
| 769 |
+
"grad_norm": 0.49862802027395625,
|
| 770 |
+
"learning_rate": 1.9304652963070868e-05,
|
| 771 |
+
"loss": 0.4925,
|
| 772 |
+
"step": 109
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.6309751434034416,
|
| 776 |
+
"grad_norm": 0.5410083146463058,
|
| 777 |
+
"learning_rate": 1.9279902461650866e-05,
|
| 778 |
+
"loss": 0.5355,
|
| 779 |
+
"step": 110
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.6367112810707457,
|
| 783 |
+
"grad_norm": 0.529469928044582,
|
| 784 |
+
"learning_rate": 1.925473557449106e-05,
|
| 785 |
+
"loss": 0.5368,
|
| 786 |
+
"step": 111
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.6424474187380497,
|
| 790 |
+
"grad_norm": 0.5510618931046878,
|
| 791 |
+
"learning_rate": 1.9229153430820232e-05,
|
| 792 |
+
"loss": 0.5431,
|
| 793 |
+
"step": 112
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.6481835564053537,
|
| 797 |
+
"grad_norm": 0.5320575572101697,
|
| 798 |
+
"learning_rate": 1.920315717849956e-05,
|
| 799 |
+
"loss": 0.5043,
|
| 800 |
+
"step": 113
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.6539196940726577,
|
| 804 |
+
"grad_norm": 0.573924225545293,
|
| 805 |
+
"learning_rate": 1.917674798397113e-05,
|
| 806 |
+
"loss": 0.5187,
|
| 807 |
+
"step": 114
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.6596558317399618,
|
| 811 |
+
"grad_norm": 0.5581615340473121,
|
| 812 |
+
"learning_rate": 1.914992703220559e-05,
|
| 813 |
+
"loss": 0.5333,
|
| 814 |
+
"step": 115
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.6653919694072657,
|
| 818 |
+
"grad_norm": 0.5234992447081256,
|
| 819 |
+
"learning_rate": 1.9122695526648968e-05,
|
| 820 |
+
"loss": 0.4942,
|
| 821 |
+
"step": 116
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.6711281070745698,
|
| 825 |
+
"grad_norm": 0.5885953422973085,
|
| 826 |
+
"learning_rate": 1.9095054689168707e-05,
|
| 827 |
+
"loss": 0.5195,
|
| 828 |
+
"step": 117
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.6768642447418738,
|
| 832 |
+
"grad_norm": 0.5297830313866135,
|
| 833 |
+
"learning_rate": 1.9067005759998797e-05,
|
| 834 |
+
"loss": 0.5423,
|
| 835 |
+
"step": 118
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.6826003824091779,
|
| 839 |
+
"grad_norm": 0.5428666103723754,
|
| 840 |
+
"learning_rate": 1.903854999768417e-05,
|
| 841 |
+
"loss": 0.5347,
|
| 842 |
+
"step": 119
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.6883365200764818,
|
| 846 |
+
"grad_norm": 0.6320535906206065,
|
| 847 |
+
"learning_rate": 1.900968867902419e-05,
|
| 848 |
+
"loss": 0.5044,
|
| 849 |
+
"step": 120
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.6940726577437859,
|
| 853 |
+
"grad_norm": 0.5752079115380615,
|
| 854 |
+
"learning_rate": 1.8980423099015402e-05,
|
| 855 |
+
"loss": 0.5411,
|
| 856 |
+
"step": 121
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.6998087954110899,
|
| 860 |
+
"grad_norm": 0.5816603517245035,
|
| 861 |
+
"learning_rate": 1.8950754570793384e-05,
|
| 862 |
+
"loss": 0.4981,
|
| 863 |
+
"step": 122
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.7055449330783938,
|
| 867 |
+
"grad_norm": 0.5761254539263562,
|
| 868 |
+
"learning_rate": 1.8920684425573865e-05,
|
| 869 |
+
"loss": 0.5067,
|
| 870 |
+
"step": 123
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.7112810707456979,
|
| 874 |
+
"grad_norm": 0.5639138995670413,
|
| 875 |
+
"learning_rate": 1.8890214012592977e-05,
|
| 876 |
+
"loss": 0.5475,
|
| 877 |
+
"step": 124
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.7170172084130019,
|
| 881 |
+
"grad_norm": 0.5486937201510219,
|
| 882 |
+
"learning_rate": 1.88593446990467e-05,
|
| 883 |
+
"loss": 0.5377,
|
| 884 |
+
"step": 125
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.722753346080306,
|
| 888 |
+
"grad_norm": 0.5333390387558036,
|
| 889 |
+
"learning_rate": 1.8828077870029554e-05,
|
| 890 |
+
"loss": 0.5067,
|
| 891 |
+
"step": 126
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.7284894837476099,
|
| 895 |
+
"grad_norm": 0.5407197244868929,
|
| 896 |
+
"learning_rate": 1.8796414928472417e-05,
|
| 897 |
+
"loss": 0.5061,
|
| 898 |
+
"step": 127
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.734225621414914,
|
| 902 |
+
"grad_norm": 0.5094464024381937,
|
| 903 |
+
"learning_rate": 1.876435729507959e-05,
|
| 904 |
+
"loss": 0.5222,
|
| 905 |
+
"step": 128
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.739961759082218,
|
| 909 |
+
"grad_norm": 0.5448472035851418,
|
| 910 |
+
"learning_rate": 1.873190640826505e-05,
|
| 911 |
+
"loss": 0.5283,
|
| 912 |
+
"step": 129
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.745697896749522,
|
| 916 |
+
"grad_norm": 0.5221064620617264,
|
| 917 |
+
"learning_rate": 1.8699063724087905e-05,
|
| 918 |
+
"loss": 0.533,
|
| 919 |
+
"step": 130
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.751434034416826,
|
| 923 |
+
"grad_norm": 0.594443241675541,
|
| 924 |
+
"learning_rate": 1.8665830716187064e-05,
|
| 925 |
+
"loss": 0.4986,
|
| 926 |
+
"step": 131
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.7571701720841301,
|
| 930 |
+
"grad_norm": 0.5508840932918168,
|
| 931 |
+
"learning_rate": 1.8632208875715122e-05,
|
| 932 |
+
"loss": 0.5052,
|
| 933 |
+
"step": 132
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.762906309751434,
|
| 937 |
+
"grad_norm": 0.6745209149751551,
|
| 938 |
+
"learning_rate": 1.8598199711271433e-05,
|
| 939 |
+
"loss": 0.4989,
|
| 940 |
+
"step": 133
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.768642447418738,
|
| 944 |
+
"grad_norm": 0.5833244098215166,
|
| 945 |
+
"learning_rate": 1.856380474883444e-05,
|
| 946 |
+
"loss": 0.4998,
|
| 947 |
+
"step": 134
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.7743785850860421,
|
| 951 |
+
"grad_norm": 0.6164702485478607,
|
| 952 |
+
"learning_rate": 1.85290255316932e-05,
|
| 953 |
+
"loss": 0.496,
|
| 954 |
+
"step": 135
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.780114722753346,
|
| 958 |
+
"grad_norm": 0.6451445851922506,
|
| 959 |
+
"learning_rate": 1.8493863620378123e-05,
|
| 960 |
+
"loss": 0.5092,
|
| 961 |
+
"step": 136
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.7858508604206501,
|
| 965 |
+
"grad_norm": 0.49991490314972736,
|
| 966 |
+
"learning_rate": 1.8458320592590976e-05,
|
| 967 |
+
"loss": 0.4953,
|
| 968 |
+
"step": 137
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.7915869980879541,
|
| 972 |
+
"grad_norm": 0.656124424238937,
|
| 973 |
+
"learning_rate": 1.8422398043134068e-05,
|
| 974 |
+
"loss": 0.5071,
|
| 975 |
+
"step": 138
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.7973231357552581,
|
| 979 |
+
"grad_norm": 0.6566346424212979,
|
| 980 |
+
"learning_rate": 1.8386097583838714e-05,
|
| 981 |
+
"loss": 0.5308,
|
| 982 |
+
"step": 139
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.8030592734225621,
|
| 986 |
+
"grad_norm": 0.6522233157913299,
|
| 987 |
+
"learning_rate": 1.834942084349289e-05,
|
| 988 |
+
"loss": 0.5547,
|
| 989 |
+
"step": 140
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.8087954110898662,
|
| 993 |
+
"grad_norm": 0.6353529825016769,
|
| 994 |
+
"learning_rate": 1.8312369467768168e-05,
|
| 995 |
+
"loss": 0.5046,
|
| 996 |
+
"step": 141
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.8145315487571702,
|
| 1000 |
+
"grad_norm": 0.6249530767355247,
|
| 1001 |
+
"learning_rate": 1.827494511914587e-05,
|
| 1002 |
+
"loss": 0.5352,
|
| 1003 |
+
"step": 142
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.8202676864244742,
|
| 1007 |
+
"grad_norm": 0.6385861110529865,
|
| 1008 |
+
"learning_rate": 1.8237149476842472e-05,
|
| 1009 |
+
"loss": 0.5029,
|
| 1010 |
+
"step": 143
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.8260038240917782,
|
| 1014 |
+
"grad_norm": 0.5607581636914353,
|
| 1015 |
+
"learning_rate": 1.8198984236734246e-05,
|
| 1016 |
+
"loss": 0.5287,
|
| 1017 |
+
"step": 144
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.8317399617590823,
|
| 1021 |
+
"grad_norm": 0.6261652427463332,
|
| 1022 |
+
"learning_rate": 1.8160451111281202e-05,
|
| 1023 |
+
"loss": 0.5114,
|
| 1024 |
+
"step": 145
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.8374760994263862,
|
| 1028 |
+
"grad_norm": 0.6528446431728675,
|
| 1029 |
+
"learning_rate": 1.81215518294502e-05,
|
| 1030 |
+
"loss": 0.5188,
|
| 1031 |
+
"step": 146
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.8432122370936902,
|
| 1035 |
+
"grad_norm": 0.5907898971360636,
|
| 1036 |
+
"learning_rate": 1.808228813663742e-05,
|
| 1037 |
+
"loss": 0.5028,
|
| 1038 |
+
"step": 147
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.8489483747609943,
|
| 1042 |
+
"grad_norm": 0.6694368765652683,
|
| 1043 |
+
"learning_rate": 1.8042661794590023e-05,
|
| 1044 |
+
"loss": 0.5054,
|
| 1045 |
+
"step": 148
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.8546845124282982,
|
| 1049 |
+
"grad_norm": 0.596432065572989,
|
| 1050 |
+
"learning_rate": 1.8002674581327096e-05,
|
| 1051 |
+
"loss": 0.49,
|
| 1052 |
+
"step": 149
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.8604206500956023,
|
| 1056 |
+
"grad_norm": 0.666622846786942,
|
| 1057 |
+
"learning_rate": 1.7962328291059886e-05,
|
| 1058 |
+
"loss": 0.5096,
|
| 1059 |
+
"step": 150
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.8661567877629063,
|
| 1063 |
+
"grad_norm": 0.673682475957719,
|
| 1064 |
+
"learning_rate": 1.7921624734111292e-05,
|
| 1065 |
+
"loss": 0.5218,
|
| 1066 |
+
"step": 151
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.8718929254302104,
|
| 1070 |
+
"grad_norm": 0.5462997037842608,
|
| 1071 |
+
"learning_rate": 1.7880565736834642e-05,
|
| 1072 |
+
"loss": 0.4693,
|
| 1073 |
+
"step": 152
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.8776290630975143,
|
| 1077 |
+
"grad_norm": 0.5908658328573468,
|
| 1078 |
+
"learning_rate": 1.783915314153172e-05,
|
| 1079 |
+
"loss": 0.5269,
|
| 1080 |
+
"step": 153
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.8833652007648184,
|
| 1084 |
+
"grad_norm": 0.6605932505871055,
|
| 1085 |
+
"learning_rate": 1.7797388806370132e-05,
|
| 1086 |
+
"loss": 0.4939,
|
| 1087 |
+
"step": 154
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.8891013384321224,
|
| 1091 |
+
"grad_norm": 0.586405871875466,
|
| 1092 |
+
"learning_rate": 1.7755274605299924e-05,
|
| 1093 |
+
"loss": 0.512,
|
| 1094 |
+
"step": 155
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.8948374760994264,
|
| 1098 |
+
"grad_norm": 0.6667799308198323,
|
| 1099 |
+
"learning_rate": 1.7712812427969485e-05,
|
| 1100 |
+
"loss": 0.5177,
|
| 1101 |
+
"step": 156
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.9005736137667304,
|
| 1105 |
+
"grad_norm": 0.5579157809901777,
|
| 1106 |
+
"learning_rate": 1.7670004179640773e-05,
|
| 1107 |
+
"loss": 0.4961,
|
| 1108 |
+
"step": 157
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.9063097514340345,
|
| 1112 |
+
"grad_norm": 0.6568561151022876,
|
| 1113 |
+
"learning_rate": 1.762685178110382e-05,
|
| 1114 |
+
"loss": 0.5299,
|
| 1115 |
+
"step": 158
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.9120458891013384,
|
| 1119 |
+
"grad_norm": 0.5190724568456453,
|
| 1120 |
+
"learning_rate": 1.7583357168590552e-05,
|
| 1121 |
+
"loss": 0.496,
|
| 1122 |
+
"step": 159
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.9177820267686424,
|
| 1126 |
+
"grad_norm": 0.6188233552623875,
|
| 1127 |
+
"learning_rate": 1.75395222936879e-05,
|
| 1128 |
+
"loss": 0.5033,
|
| 1129 |
+
"step": 160
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.9235181644359465,
|
| 1133 |
+
"grad_norm": 0.48950716483283546,
|
| 1134 |
+
"learning_rate": 1.7495349123250242e-05,
|
| 1135 |
+
"loss": 0.5201,
|
| 1136 |
+
"step": 161
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.9292543021032504,
|
| 1140 |
+
"grad_norm": 0.6250127202556996,
|
| 1141 |
+
"learning_rate": 1.7450839639311164e-05,
|
| 1142 |
+
"loss": 0.4674,
|
| 1143 |
+
"step": 162
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.9349904397705545,
|
| 1147 |
+
"grad_norm": 0.5109679077677434,
|
| 1148 |
+
"learning_rate": 1.7405995838994495e-05,
|
| 1149 |
+
"loss": 0.5182,
|
| 1150 |
+
"step": 163
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.9407265774378585,
|
| 1154 |
+
"grad_norm": 0.5858835531348051,
|
| 1155 |
+
"learning_rate": 1.7360819734424718e-05,
|
| 1156 |
+
"loss": 0.518,
|
| 1157 |
+
"step": 164
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.9464627151051626,
|
| 1161 |
+
"grad_norm": 0.5788419193604027,
|
| 1162 |
+
"learning_rate": 1.731531335263669e-05,
|
| 1163 |
+
"loss": 0.5223,
|
| 1164 |
+
"step": 165
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.9521988527724665,
|
| 1168 |
+
"grad_norm": 0.5789436218944392,
|
| 1169 |
+
"learning_rate": 1.7269478735484682e-05,
|
| 1170 |
+
"loss": 0.4855,
|
| 1171 |
+
"step": 166
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.9579349904397706,
|
| 1175 |
+
"grad_norm": 0.5208960460250472,
|
| 1176 |
+
"learning_rate": 1.7223317939550756e-05,
|
| 1177 |
+
"loss": 0.4881,
|
| 1178 |
+
"step": 167
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.9636711281070746,
|
| 1182 |
+
"grad_norm": 0.5099943351497956,
|
| 1183 |
+
"learning_rate": 1.7176833036052495e-05,
|
| 1184 |
+
"loss": 0.4835,
|
| 1185 |
+
"step": 168
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.9694072657743786,
|
| 1189 |
+
"grad_norm": 0.5897069652807404,
|
| 1190 |
+
"learning_rate": 1.713002611075007e-05,
|
| 1191 |
+
"loss": 0.507,
|
| 1192 |
+
"step": 169
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.9751434034416826,
|
| 1196 |
+
"grad_norm": 0.564216052819639,
|
| 1197 |
+
"learning_rate": 1.708289926385265e-05,
|
| 1198 |
+
"loss": 0.4979,
|
| 1199 |
+
"step": 170
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.9808795411089866,
|
| 1203 |
+
"grad_norm": 0.5326107874470021,
|
| 1204 |
+
"learning_rate": 1.703545460992416e-05,
|
| 1205 |
+
"loss": 0.5437,
|
| 1206 |
+
"step": 171
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.9866156787762906,
|
| 1210 |
+
"grad_norm": 0.5578118857749461,
|
| 1211 |
+
"learning_rate": 1.698769427778842e-05,
|
| 1212 |
+
"loss": 0.497,
|
| 1213 |
+
"step": 172
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.9923518164435946,
|
| 1217 |
+
"grad_norm": 0.5684954816297212,
|
| 1218 |
+
"learning_rate": 1.693962041043359e-05,
|
| 1219 |
+
"loss": 0.5283,
|
| 1220 |
+
"step": 173
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.9980879541108987,
|
| 1224 |
+
"grad_norm": 0.5422375671625977,
|
| 1225 |
+
"learning_rate": 1.6891235164916066e-05,
|
| 1226 |
+
"loss": 0.5188,
|
| 1227 |
+
"step": 174
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.0038240917782026,
|
| 1231 |
+
"grad_norm": 0.8788446159577279,
|
| 1232 |
+
"learning_rate": 1.684254071226364e-05,
|
| 1233 |
+
"loss": 0.726,
|
| 1234 |
+
"step": 175
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.0095602294455066,
|
| 1238 |
+
"grad_norm": 0.6280093992507375,
|
| 1239 |
+
"learning_rate": 1.679353923737813e-05,
|
| 1240 |
+
"loss": 0.4404,
|
| 1241 |
+
"step": 176
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.0152963671128108,
|
| 1245 |
+
"grad_norm": 0.6259953429427509,
|
| 1246 |
+
"learning_rate": 1.674423293893731e-05,
|
| 1247 |
+
"loss": 0.4864,
|
| 1248 |
+
"step": 177
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.0210325047801148,
|
| 1252 |
+
"grad_norm": 0.5979650245228064,
|
| 1253 |
+
"learning_rate": 1.6694624029296288e-05,
|
| 1254 |
+
"loss": 0.4239,
|
| 1255 |
+
"step": 178
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.0267686424474187,
|
| 1259 |
+
"grad_norm": 0.6847243364062234,
|
| 1260 |
+
"learning_rate": 1.664471473438822e-05,
|
| 1261 |
+
"loss": 0.419,
|
| 1262 |
+
"step": 179
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.0325047801147227,
|
| 1266 |
+
"grad_norm": 0.5744272580087806,
|
| 1267 |
+
"learning_rate": 1.6594507293624426e-05,
|
| 1268 |
+
"loss": 0.4517,
|
| 1269 |
+
"step": 180
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.0382409177820269,
|
| 1273 |
+
"grad_norm": 0.5902807866572417,
|
| 1274 |
+
"learning_rate": 1.6544003959793925e-05,
|
| 1275 |
+
"loss": 0.3741,
|
| 1276 |
+
"step": 181
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 1.0439770554493308,
|
| 1280 |
+
"grad_norm": 0.6855325601836614,
|
| 1281 |
+
"learning_rate": 1.6493206998962353e-05,
|
| 1282 |
+
"loss": 0.4579,
|
| 1283 |
+
"step": 182
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 1.0497131931166348,
|
| 1287 |
+
"grad_norm": 0.5940167639272883,
|
| 1288 |
+
"learning_rate": 1.6442118690370272e-05,
|
| 1289 |
+
"loss": 0.4572,
|
| 1290 |
+
"step": 183
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.0554493307839388,
|
| 1294 |
+
"grad_norm": 0.5956164862309621,
|
| 1295 |
+
"learning_rate": 1.639074132633091e-05,
|
| 1296 |
+
"loss": 0.4421,
|
| 1297 |
+
"step": 184
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.0611854684512427,
|
| 1301 |
+
"grad_norm": 0.610351381314269,
|
| 1302 |
+
"learning_rate": 1.6339077212127294e-05,
|
| 1303 |
+
"loss": 0.4006,
|
| 1304 |
+
"step": 185
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 1.066921606118547,
|
| 1308 |
+
"grad_norm": 0.6717976632223687,
|
| 1309 |
+
"learning_rate": 1.628712866590885e-05,
|
| 1310 |
+
"loss": 0.4775,
|
| 1311 |
+
"step": 186
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.0726577437858509,
|
| 1315 |
+
"grad_norm": 0.6986452772586962,
|
| 1316 |
+
"learning_rate": 1.6234898018587336e-05,
|
| 1317 |
+
"loss": 0.4448,
|
| 1318 |
+
"step": 187
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 1.0783938814531548,
|
| 1322 |
+
"grad_norm": 0.6297000888622394,
|
| 1323 |
+
"learning_rate": 1.6182387613732295e-05,
|
| 1324 |
+
"loss": 0.4897,
|
| 1325 |
+
"step": 188
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 1.0841300191204588,
|
| 1329 |
+
"grad_norm": 0.4850479578653087,
|
| 1330 |
+
"learning_rate": 1.6129599807465873e-05,
|
| 1331 |
+
"loss": 0.3727,
|
| 1332 |
+
"step": 189
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 1.089866156787763,
|
| 1336 |
+
"grad_norm": 0.6942735667602296,
|
| 1337 |
+
"learning_rate": 1.607653696835713e-05,
|
| 1338 |
+
"loss": 0.5011,
|
| 1339 |
+
"step": 190
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 1.095602294455067,
|
| 1343 |
+
"grad_norm": 0.546352054633847,
|
| 1344 |
+
"learning_rate": 1.602320147731573e-05,
|
| 1345 |
+
"loss": 0.4083,
|
| 1346 |
+
"step": 191
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 1.101338432122371,
|
| 1350 |
+
"grad_norm": 0.497459437509141,
|
| 1351 |
+
"learning_rate": 1.596959572748514e-05,
|
| 1352 |
+
"loss": 0.4053,
|
| 1353 |
+
"step": 192
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.107074569789675,
|
| 1357 |
+
"grad_norm": 0.5915697095267101,
|
| 1358 |
+
"learning_rate": 1.5915722124135227e-05,
|
| 1359 |
+
"loss": 0.4421,
|
| 1360 |
+
"step": 193
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 1.1128107074569789,
|
| 1364 |
+
"grad_norm": 0.4822188270663418,
|
| 1365 |
+
"learning_rate": 1.586158308455435e-05,
|
| 1366 |
+
"loss": 0.4458,
|
| 1367 |
+
"step": 194
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 1.118546845124283,
|
| 1371 |
+
"grad_norm": 0.586482650803536,
|
| 1372 |
+
"learning_rate": 1.580718103794089e-05,
|
| 1373 |
+
"loss": 0.4402,
|
| 1374 |
+
"step": 195
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.124282982791587,
|
| 1378 |
+
"grad_norm": 0.5577297526984457,
|
| 1379 |
+
"learning_rate": 1.5752518425294258e-05,
|
| 1380 |
+
"loss": 0.3781,
|
| 1381 |
+
"step": 196
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 1.130019120458891,
|
| 1385 |
+
"grad_norm": 0.5870354491149415,
|
| 1386 |
+
"learning_rate": 1.569759769930537e-05,
|
| 1387 |
+
"loss": 0.5061,
|
| 1388 |
+
"step": 197
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 1.135755258126195,
|
| 1392 |
+
"grad_norm": 0.6371195649882143,
|
| 1393 |
+
"learning_rate": 1.564242132424657e-05,
|
| 1394 |
+
"loss": 0.4414,
|
| 1395 |
+
"step": 198
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.1414913957934991,
|
| 1399 |
+
"grad_norm": 0.4589988573509597,
|
| 1400 |
+
"learning_rate": 1.5586991775861103e-05,
|
| 1401 |
+
"loss": 0.3934,
|
| 1402 |
+
"step": 199
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 1.147227533460803,
|
| 1406 |
+
"grad_norm": 0.5493331894713434,
|
| 1407 |
+
"learning_rate": 1.5531311541251995e-05,
|
| 1408 |
+
"loss": 0.4114,
|
| 1409 |
+
"step": 200
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 1.152963671128107,
|
| 1413 |
+
"grad_norm": 0.5138358962197389,
|
| 1414 |
+
"learning_rate": 1.5475383118770473e-05,
|
| 1415 |
+
"loss": 0.4159,
|
| 1416 |
+
"step": 201
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 1.158699808795411,
|
| 1420 |
+
"grad_norm": 0.6607382125199128,
|
| 1421 |
+
"learning_rate": 1.5419209017903855e-05,
|
| 1422 |
+
"loss": 0.4859,
|
| 1423 |
+
"step": 202
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 1.1644359464627152,
|
| 1427 |
+
"grad_norm": 0.476778446402774,
|
| 1428 |
+
"learning_rate": 1.536279175916296e-05,
|
| 1429 |
+
"loss": 0.402,
|
| 1430 |
+
"step": 203
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 1.1701720841300192,
|
| 1434 |
+
"grad_norm": 0.662594438635697,
|
| 1435 |
+
"learning_rate": 1.5306133873969008e-05,
|
| 1436 |
+
"loss": 0.4958,
|
| 1437 |
+
"step": 204
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 1.1759082217973231,
|
| 1441 |
+
"grad_norm": 0.49342011546672193,
|
| 1442 |
+
"learning_rate": 1.5249237904540041e-05,
|
| 1443 |
+
"loss": 0.4438,
|
| 1444 |
+
"step": 205
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 1.181644359464627,
|
| 1448 |
+
"grad_norm": 0.56704289045039,
|
| 1449 |
+
"learning_rate": 1.519210640377685e-05,
|
| 1450 |
+
"loss": 0.4239,
|
| 1451 |
+
"step": 206
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 1.1873804971319313,
|
| 1455 |
+
"grad_norm": 0.5047410935107941,
|
| 1456 |
+
"learning_rate": 1.513474193514842e-05,
|
| 1457 |
+
"loss": 0.4288,
|
| 1458 |
+
"step": 207
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 1.1931166347992352,
|
| 1462 |
+
"grad_norm": 0.5926076517541518,
|
| 1463 |
+
"learning_rate": 1.5077147072576932e-05,
|
| 1464 |
+
"loss": 0.4277,
|
| 1465 |
+
"step": 208
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 1.1988527724665392,
|
| 1469 |
+
"grad_norm": 0.49311177861969074,
|
| 1470 |
+
"learning_rate": 1.5019324400322244e-05,
|
| 1471 |
+
"loss": 0.4359,
|
| 1472 |
+
"step": 209
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 1.2045889101338432,
|
| 1476 |
+
"grad_norm": 0.5662399638263612,
|
| 1477 |
+
"learning_rate": 1.4961276512865954e-05,
|
| 1478 |
+
"loss": 0.4269,
|
| 1479 |
+
"step": 210
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 1.2103250478011471,
|
| 1483 |
+
"grad_norm": 0.4759114943666039,
|
| 1484 |
+
"learning_rate": 1.4903006014794983e-05,
|
| 1485 |
+
"loss": 0.4509,
|
| 1486 |
+
"step": 211
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 1.2160611854684513,
|
| 1490 |
+
"grad_norm": 0.6159634050140991,
|
| 1491 |
+
"learning_rate": 1.4844515520684703e-05,
|
| 1492 |
+
"loss": 0.443,
|
| 1493 |
+
"step": 212
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 1.2217973231357553,
|
| 1497 |
+
"grad_norm": 0.5145704198069356,
|
| 1498 |
+
"learning_rate": 1.4785807654981627e-05,
|
| 1499 |
+
"loss": 0.4079,
|
| 1500 |
+
"step": 213
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 1.2275334608030593,
|
| 1504 |
+
"grad_norm": 0.5006839310785222,
|
| 1505 |
+
"learning_rate": 1.4726885051885654e-05,
|
| 1506 |
+
"loss": 0.4102,
|
| 1507 |
+
"step": 214
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 1.2332695984703632,
|
| 1511 |
+
"grad_norm": 0.6434261075702461,
|
| 1512 |
+
"learning_rate": 1.4667750355231863e-05,
|
| 1513 |
+
"loss": 0.4577,
|
| 1514 |
+
"step": 215
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 1.2390057361376674,
|
| 1518 |
+
"grad_norm": 0.49583426427426724,
|
| 1519 |
+
"learning_rate": 1.4608406218371894e-05,
|
| 1520 |
+
"loss": 0.4538,
|
| 1521 |
+
"step": 216
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 1.2447418738049714,
|
| 1525 |
+
"grad_norm": 0.5367458222855271,
|
| 1526 |
+
"learning_rate": 1.4548855304054888e-05,
|
| 1527 |
+
"loss": 0.3844,
|
| 1528 |
+
"step": 217
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 1.2504780114722753,
|
| 1532 |
+
"grad_norm": 0.5081216759568395,
|
| 1533 |
+
"learning_rate": 1.4489100284308018e-05,
|
| 1534 |
+
"loss": 0.3947,
|
| 1535 |
+
"step": 218
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 1.2562141491395793,
|
| 1539 |
+
"grad_norm": 0.585650284147855,
|
| 1540 |
+
"learning_rate": 1.4429143840316586e-05,
|
| 1541 |
+
"loss": 0.4645,
|
| 1542 |
+
"step": 219
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 1.2619502868068833,
|
| 1546 |
+
"grad_norm": 0.5171763518723317,
|
| 1547 |
+
"learning_rate": 1.4368988662303733e-05,
|
| 1548 |
+
"loss": 0.452,
|
| 1549 |
+
"step": 220
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 1.2676864244741874,
|
| 1553 |
+
"grad_norm": 0.48039143238596627,
|
| 1554 |
+
"learning_rate": 1.4308637449409705e-05,
|
| 1555 |
+
"loss": 0.441,
|
| 1556 |
+
"step": 221
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 1.2734225621414914,
|
| 1560 |
+
"grad_norm": 0.5684128265489994,
|
| 1561 |
+
"learning_rate": 1.4248092909570774e-05,
|
| 1562 |
+
"loss": 0.457,
|
| 1563 |
+
"step": 222
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 1.2791586998087954,
|
| 1567 |
+
"grad_norm": 0.6202786095376327,
|
| 1568 |
+
"learning_rate": 1.4187357759397716e-05,
|
| 1569 |
+
"loss": 0.474,
|
| 1570 |
+
"step": 223
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 1.2848948374760996,
|
| 1574 |
+
"grad_norm": 0.5132214354576209,
|
| 1575 |
+
"learning_rate": 1.4126434724053915e-05,
|
| 1576 |
+
"loss": 0.4418,
|
| 1577 |
+
"step": 224
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 1.2906309751434035,
|
| 1581 |
+
"grad_norm": 0.47833838706648946,
|
| 1582 |
+
"learning_rate": 1.4065326537133094e-05,
|
| 1583 |
+
"loss": 0.4143,
|
| 1584 |
+
"step": 225
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 1.2963671128107075,
|
| 1588 |
+
"grad_norm": 0.5126041137931057,
|
| 1589 |
+
"learning_rate": 1.4004035940536671e-05,
|
| 1590 |
+
"loss": 0.4301,
|
| 1591 |
+
"step": 226
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 1.3021032504780115,
|
| 1595 |
+
"grad_norm": 0.4906072135255895,
|
| 1596 |
+
"learning_rate": 1.39425656843507e-05,
|
| 1597 |
+
"loss": 0.439,
|
| 1598 |
+
"step": 227
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 1.3078393881453154,
|
| 1602 |
+
"grad_norm": 0.5366969388908639,
|
| 1603 |
+
"learning_rate": 1.3880918526722497e-05,
|
| 1604 |
+
"loss": 0.4653,
|
| 1605 |
+
"step": 228
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 1.3135755258126194,
|
| 1609 |
+
"grad_norm": 0.5481791341858512,
|
| 1610 |
+
"learning_rate": 1.3819097233736888e-05,
|
| 1611 |
+
"loss": 0.4895,
|
| 1612 |
+
"step": 229
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 1.3193116634799236,
|
| 1616 |
+
"grad_norm": 0.46387604676620864,
|
| 1617 |
+
"learning_rate": 1.3757104579292082e-05,
|
| 1618 |
+
"loss": 0.407,
|
| 1619 |
+
"step": 230
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 1.3250478011472275,
|
| 1623 |
+
"grad_norm": 0.47558528540904305,
|
| 1624 |
+
"learning_rate": 1.3694943344975214e-05,
|
| 1625 |
+
"loss": 0.3897,
|
| 1626 |
+
"step": 231
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 1.3307839388145315,
|
| 1630 |
+
"grad_norm": 0.4776725124785085,
|
| 1631 |
+
"learning_rate": 1.3632616319937522e-05,
|
| 1632 |
+
"loss": 0.4569,
|
| 1633 |
+
"step": 232
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 1.3365200764818357,
|
| 1637 |
+
"grad_norm": 0.5538110118526167,
|
| 1638 |
+
"learning_rate": 1.3570126300769233e-05,
|
| 1639 |
+
"loss": 0.4563,
|
| 1640 |
+
"step": 233
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 1.3422562141491396,
|
| 1644 |
+
"grad_norm": 0.463107688708888,
|
| 1645 |
+
"learning_rate": 1.3507476091374042e-05,
|
| 1646 |
+
"loss": 0.454,
|
| 1647 |
+
"step": 234
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 1.3479923518164436,
|
| 1651 |
+
"grad_norm": 0.5020783938759849,
|
| 1652 |
+
"learning_rate": 1.344466850284333e-05,
|
| 1653 |
+
"loss": 0.4314,
|
| 1654 |
+
"step": 235
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 1.3537284894837476,
|
| 1658 |
+
"grad_norm": 0.6036785088586939,
|
| 1659 |
+
"learning_rate": 1.3381706353330015e-05,
|
| 1660 |
+
"loss": 0.4783,
|
| 1661 |
+
"step": 236
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 1.3594646271510515,
|
| 1665 |
+
"grad_norm": 0.45597006215248087,
|
| 1666 |
+
"learning_rate": 1.331859246792211e-05,
|
| 1667 |
+
"loss": 0.3658,
|
| 1668 |
+
"step": 237
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 1.3652007648183555,
|
| 1672 |
+
"grad_norm": 0.6491828764472608,
|
| 1673 |
+
"learning_rate": 1.3255329678515959e-05,
|
| 1674 |
+
"loss": 0.456,
|
| 1675 |
+
"step": 238
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 1.3709369024856597,
|
| 1679 |
+
"grad_norm": 0.5148839976999712,
|
| 1680 |
+
"learning_rate": 1.3191920823689178e-05,
|
| 1681 |
+
"loss": 0.4056,
|
| 1682 |
+
"step": 239
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 1.3766730401529637,
|
| 1686 |
+
"grad_norm": 0.5798134301190937,
|
| 1687 |
+
"learning_rate": 1.3128368748573272e-05,
|
| 1688 |
+
"loss": 0.4938,
|
| 1689 |
+
"step": 240
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 1.3824091778202676,
|
| 1693 |
+
"grad_norm": 0.5118762044788047,
|
| 1694 |
+
"learning_rate": 1.3064676304726001e-05,
|
| 1695 |
+
"loss": 0.4354,
|
| 1696 |
+
"step": 241
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 1.3881453154875718,
|
| 1700 |
+
"grad_norm": 0.5579407020665195,
|
| 1701 |
+
"learning_rate": 1.300084635000341e-05,
|
| 1702 |
+
"loss": 0.4009,
|
| 1703 |
+
"step": 242
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 1.3938814531548758,
|
| 1707 |
+
"grad_norm": 0.47914049556136357,
|
| 1708 |
+
"learning_rate": 1.2936881748431601e-05,
|
| 1709 |
+
"loss": 0.4139,
|
| 1710 |
+
"step": 243
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 1.3996175908221797,
|
| 1714 |
+
"grad_norm": 0.6781741684919476,
|
| 1715 |
+
"learning_rate": 1.287278537007824e-05,
|
| 1716 |
+
"loss": 0.4831,
|
| 1717 |
+
"step": 244
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 1.4053537284894837,
|
| 1721 |
+
"grad_norm": 0.47111455679765774,
|
| 1722 |
+
"learning_rate": 1.280856009092376e-05,
|
| 1723 |
+
"loss": 0.4599,
|
| 1724 |
+
"step": 245
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 1.4110898661567877,
|
| 1728 |
+
"grad_norm": 0.46738967278396626,
|
| 1729 |
+
"learning_rate": 1.2744208792732324e-05,
|
| 1730 |
+
"loss": 0.4387,
|
| 1731 |
+
"step": 246
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 1.4168260038240919,
|
| 1735 |
+
"grad_norm": 0.5304710050520607,
|
| 1736 |
+
"learning_rate": 1.267973436292253e-05,
|
| 1737 |
+
"loss": 0.4513,
|
| 1738 |
+
"step": 247
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 1.4225621414913958,
|
| 1742 |
+
"grad_norm": 0.43485659330912546,
|
| 1743 |
+
"learning_rate": 1.2615139694437835e-05,
|
| 1744 |
+
"loss": 0.4107,
|
| 1745 |
+
"step": 248
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 1.4282982791586998,
|
| 1749 |
+
"grad_norm": 0.4874553497224479,
|
| 1750 |
+
"learning_rate": 1.2550427685616767e-05,
|
| 1751 |
+
"loss": 0.4504,
|
| 1752 |
+
"step": 249
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 1.4340344168260037,
|
| 1756 |
+
"grad_norm": 0.47534151482068415,
|
| 1757 |
+
"learning_rate": 1.2485601240062868e-05,
|
| 1758 |
+
"loss": 0.4142,
|
| 1759 |
+
"step": 250
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 1.439770554493308,
|
| 1763 |
+
"grad_norm": 0.49136255110854465,
|
| 1764 |
+
"learning_rate": 1.2420663266514419e-05,
|
| 1765 |
+
"loss": 0.4523,
|
| 1766 |
+
"step": 251
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 1.445506692160612,
|
| 1770 |
+
"grad_norm": 0.4859920796473984,
|
| 1771 |
+
"learning_rate": 1.2355616678713909e-05,
|
| 1772 |
+
"loss": 0.4543,
|
| 1773 |
+
"step": 252
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 1.4512428298279159,
|
| 1777 |
+
"grad_norm": 0.4106107196876608,
|
| 1778 |
+
"learning_rate": 1.229046439527732e-05,
|
| 1779 |
+
"loss": 0.3931,
|
| 1780 |
+
"step": 253
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 1.4569789674952198,
|
| 1784 |
+
"grad_norm": 0.4971544355198163,
|
| 1785 |
+
"learning_rate": 1.2225209339563144e-05,
|
| 1786 |
+
"loss": 0.44,
|
| 1787 |
+
"step": 254
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 1.4627151051625238,
|
| 1791 |
+
"grad_norm": 0.4521844159684417,
|
| 1792 |
+
"learning_rate": 1.2159854439541245e-05,
|
| 1793 |
+
"loss": 0.4551,
|
| 1794 |
+
"step": 255
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 1.468451242829828,
|
| 1798 |
+
"grad_norm": 0.45723395845362985,
|
| 1799 |
+
"learning_rate": 1.2094402627661447e-05,
|
| 1800 |
+
"loss": 0.443,
|
| 1801 |
+
"step": 256
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 1.474187380497132,
|
| 1805 |
+
"grad_norm": 0.4826661110709425,
|
| 1806 |
+
"learning_rate": 1.2028856840721975e-05,
|
| 1807 |
+
"loss": 0.4656,
|
| 1808 |
+
"step": 257
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 1.479923518164436,
|
| 1812 |
+
"grad_norm": 0.39565718965895136,
|
| 1813 |
+
"learning_rate": 1.1963220019737691e-05,
|
| 1814 |
+
"loss": 0.4234,
|
| 1815 |
+
"step": 258
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 1.48565965583174,
|
| 1819 |
+
"grad_norm": 0.47431335698220256,
|
| 1820 |
+
"learning_rate": 1.1897495109808108e-05,
|
| 1821 |
+
"loss": 0.4543,
|
| 1822 |
+
"step": 259
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 1.491395793499044,
|
| 1826 |
+
"grad_norm": 0.4507720979158589,
|
| 1827 |
+
"learning_rate": 1.1831685059985263e-05,
|
| 1828 |
+
"loss": 0.4546,
|
| 1829 |
+
"step": 260
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 1.497131931166348,
|
| 1833 |
+
"grad_norm": 0.40102046460904306,
|
| 1834 |
+
"learning_rate": 1.1765792823141385e-05,
|
| 1835 |
+
"loss": 0.4111,
|
| 1836 |
+
"step": 261
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 1.502868068833652,
|
| 1840 |
+
"grad_norm": 0.4589104612978405,
|
| 1841 |
+
"learning_rate": 1.169982135583641e-05,
|
| 1842 |
+
"loss": 0.4134,
|
| 1843 |
+
"step": 262
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 1.508604206500956,
|
| 1847 |
+
"grad_norm": 0.4683338597962924,
|
| 1848 |
+
"learning_rate": 1.1633773618185302e-05,
|
| 1849 |
+
"loss": 0.4397,
|
| 1850 |
+
"step": 263
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 1.51434034416826,
|
| 1854 |
+
"grad_norm": 0.4558640349219329,
|
| 1855 |
+
"learning_rate": 1.1567652573725263e-05,
|
| 1856 |
+
"loss": 0.4484,
|
| 1857 |
+
"step": 264
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 1.520076481835564,
|
| 1861 |
+
"grad_norm": 0.439951790644891,
|
| 1862 |
+
"learning_rate": 1.1501461189282734e-05,
|
| 1863 |
+
"loss": 0.4027,
|
| 1864 |
+
"step": 265
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 1.525812619502868,
|
| 1868 |
+
"grad_norm": 0.44164693690672613,
|
| 1869 |
+
"learning_rate": 1.1435202434840287e-05,
|
| 1870 |
+
"loss": 0.4266,
|
| 1871 |
+
"step": 266
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 1.5315487571701722,
|
| 1875 |
+
"grad_norm": 0.43551426817891264,
|
| 1876 |
+
"learning_rate": 1.136887928340336e-05,
|
| 1877 |
+
"loss": 0.4279,
|
| 1878 |
+
"step": 267
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 1.5372848948374762,
|
| 1882 |
+
"grad_norm": 0.460751533150955,
|
| 1883 |
+
"learning_rate": 1.1302494710866859e-05,
|
| 1884 |
+
"loss": 0.4179,
|
| 1885 |
+
"step": 268
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 1.5430210325047802,
|
| 1889 |
+
"grad_norm": 0.4746552779750239,
|
| 1890 |
+
"learning_rate": 1.1236051695881634e-05,
|
| 1891 |
+
"loss": 0.4769,
|
| 1892 |
+
"step": 269
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 1.5487571701720841,
|
| 1896 |
+
"grad_norm": 0.47105209380963015,
|
| 1897 |
+
"learning_rate": 1.1169553219720828e-05,
|
| 1898 |
+
"loss": 0.4599,
|
| 1899 |
+
"step": 270
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 1.554493307839388,
|
| 1903 |
+
"grad_norm": 0.4566600321680249,
|
| 1904 |
+
"learning_rate": 1.1103002266146096e-05,
|
| 1905 |
+
"loss": 0.4515,
|
| 1906 |
+
"step": 271
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 1.560229445506692,
|
| 1910 |
+
"grad_norm": 0.387915706277622,
|
| 1911 |
+
"learning_rate": 1.103640182127375e-05,
|
| 1912 |
+
"loss": 0.4265,
|
| 1913 |
+
"step": 272
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 1.565965583173996,
|
| 1917 |
+
"grad_norm": 0.44038850987015166,
|
| 1918 |
+
"learning_rate": 1.0969754873440743e-05,
|
| 1919 |
+
"loss": 0.4486,
|
| 1920 |
+
"step": 273
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 1.5717017208413002,
|
| 1924 |
+
"grad_norm": 0.46869796330160723,
|
| 1925 |
+
"learning_rate": 1.0903064413070611e-05,
|
| 1926 |
+
"loss": 0.4834,
|
| 1927 |
+
"step": 274
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 1.5774378585086042,
|
| 1931 |
+
"grad_norm": 0.40500478010238716,
|
| 1932 |
+
"learning_rate": 1.0836333432539272e-05,
|
| 1933 |
+
"loss": 0.4084,
|
| 1934 |
+
"step": 275
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 1.5831739961759084,
|
| 1938 |
+
"grad_norm": 0.39581081249275896,
|
| 1939 |
+
"learning_rate": 1.076956492604077e-05,
|
| 1940 |
+
"loss": 0.4574,
|
| 1941 |
+
"step": 276
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 1.5889101338432123,
|
| 1945 |
+
"grad_norm": 0.4621337562923899,
|
| 1946 |
+
"learning_rate": 1.070276188945293e-05,
|
| 1947 |
+
"loss": 0.44,
|
| 1948 |
+
"step": 277
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 1.5946462715105163,
|
| 1952 |
+
"grad_norm": 0.403297544236656,
|
| 1953 |
+
"learning_rate": 1.0635927320202928e-05,
|
| 1954 |
+
"loss": 0.428,
|
| 1955 |
+
"step": 278
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 1.6003824091778203,
|
| 1959 |
+
"grad_norm": 0.40429520535585933,
|
| 1960 |
+
"learning_rate": 1.0569064217132791e-05,
|
| 1961 |
+
"loss": 0.3999,
|
| 1962 |
+
"step": 279
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 1.6061185468451242,
|
| 1966 |
+
"grad_norm": 0.44992301434912374,
|
| 1967 |
+
"learning_rate": 1.0502175580364857e-05,
|
| 1968 |
+
"loss": 0.4629,
|
| 1969 |
+
"step": 280
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 1.6118546845124282,
|
| 1973 |
+
"grad_norm": 0.5317926764729383,
|
| 1974 |
+
"learning_rate": 1.0435264411167148e-05,
|
| 1975 |
+
"loss": 0.426,
|
| 1976 |
+
"step": 281
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 1.6175908221797322,
|
| 1980 |
+
"grad_norm": 0.43995618220520716,
|
| 1981 |
+
"learning_rate": 1.036833371181871e-05,
|
| 1982 |
+
"loss": 0.4835,
|
| 1983 |
+
"step": 282
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 1.6233269598470363,
|
| 1987 |
+
"grad_norm": 0.401770871391019,
|
| 1988 |
+
"learning_rate": 1.0301386485474888e-05,
|
| 1989 |
+
"loss": 0.4303,
|
| 1990 |
+
"step": 283
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 1.6290630975143403,
|
| 1994 |
+
"grad_norm": 0.3924022810984957,
|
| 1995 |
+
"learning_rate": 1.0234425736032607e-05,
|
| 1996 |
+
"loss": 0.4528,
|
| 1997 |
+
"step": 284
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 1.6347992351816445,
|
| 2001 |
+
"grad_norm": 0.43649719543388227,
|
| 2002 |
+
"learning_rate": 1.016745446799555e-05,
|
| 2003 |
+
"loss": 0.4054,
|
| 2004 |
+
"step": 285
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 1.6405353728489485,
|
| 2008 |
+
"grad_norm": 0.44190590472557323,
|
| 2009 |
+
"learning_rate": 1.010047568633938e-05,
|
| 2010 |
+
"loss": 0.4622,
|
| 2011 |
+
"step": 286
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 1.6462715105162524,
|
| 2015 |
+
"grad_norm": 0.4138119608613635,
|
| 2016 |
+
"learning_rate": 1.0033492396376879e-05,
|
| 2017 |
+
"loss": 0.4443,
|
| 2018 |
+
"step": 287
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 1.6520076481835564,
|
| 2022 |
+
"grad_norm": 0.410370717741024,
|
| 2023 |
+
"learning_rate": 9.966507603623125e-06,
|
| 2024 |
+
"loss": 0.4255,
|
| 2025 |
+
"step": 288
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 1.6577437858508604,
|
| 2029 |
+
"grad_norm": 0.3851173486609221,
|
| 2030 |
+
"learning_rate": 9.899524313660623e-06,
|
| 2031 |
+
"loss": 0.439,
|
| 2032 |
+
"step": 289
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 1.6634799235181643,
|
| 2036 |
+
"grad_norm": 0.38567722494532186,
|
| 2037 |
+
"learning_rate": 9.832545532004454e-06,
|
| 2038 |
+
"loss": 0.4339,
|
| 2039 |
+
"step": 290
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 1.6692160611854685,
|
| 2043 |
+
"grad_norm": 0.3955204462600041,
|
| 2044 |
+
"learning_rate": 9.765574263967397e-06,
|
| 2045 |
+
"loss": 0.3977,
|
| 2046 |
+
"step": 291
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 1.6749521988527725,
|
| 2050 |
+
"grad_norm": 0.42445834780458475,
|
| 2051 |
+
"learning_rate": 9.698613514525117e-06,
|
| 2052 |
+
"loss": 0.422,
|
| 2053 |
+
"step": 292
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 1.6806883365200764,
|
| 2057 |
+
"grad_norm": 0.44194756806146035,
|
| 2058 |
+
"learning_rate": 9.631666288181293e-06,
|
| 2059 |
+
"loss": 0.4903,
|
| 2060 |
+
"step": 293
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 1.6864244741873806,
|
| 2064 |
+
"grad_norm": 0.46162371183733075,
|
| 2065 |
+
"learning_rate": 9.564735588832857e-06,
|
| 2066 |
+
"loss": 0.5333,
|
| 2067 |
+
"step": 294
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 1.6921606118546846,
|
| 2071 |
+
"grad_norm": 0.4051316346724794,
|
| 2072 |
+
"learning_rate": 9.497824419635145e-06,
|
| 2073 |
+
"loss": 0.3805,
|
| 2074 |
+
"step": 295
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 1.6978967495219885,
|
| 2078 |
+
"grad_norm": 0.3960838463524368,
|
| 2079 |
+
"learning_rate": 9.430935782867214e-06,
|
| 2080 |
+
"loss": 0.4379,
|
| 2081 |
+
"step": 296
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 1.7036328871892925,
|
| 2085 |
+
"grad_norm": 0.40765156595116686,
|
| 2086 |
+
"learning_rate": 9.364072679797074e-06,
|
| 2087 |
+
"loss": 0.439,
|
| 2088 |
+
"step": 297
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 1.7093690248565965,
|
| 2092 |
+
"grad_norm": 0.4258417890684251,
|
| 2093 |
+
"learning_rate": 9.297238110547075e-06,
|
| 2094 |
+
"loss": 0.4748,
|
| 2095 |
+
"step": 298
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 1.7151051625239004,
|
| 2099 |
+
"grad_norm": 0.398962552995342,
|
| 2100 |
+
"learning_rate": 9.230435073959232e-06,
|
| 2101 |
+
"loss": 0.4381,
|
| 2102 |
+
"step": 299
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 1.7208413001912046,
|
| 2106 |
+
"grad_norm": 0.38078824772489206,
|
| 2107 |
+
"learning_rate": 9.163666567460735e-06,
|
| 2108 |
+
"loss": 0.3894,
|
| 2109 |
+
"step": 300
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 1.7265774378585086,
|
| 2113 |
+
"grad_norm": 0.4002207799336185,
|
| 2114 |
+
"learning_rate": 9.096935586929392e-06,
|
| 2115 |
+
"loss": 0.4284,
|
| 2116 |
+
"step": 301
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 1.7323135755258128,
|
| 2120 |
+
"grad_norm": 0.4385360843975406,
|
| 2121 |
+
"learning_rate": 9.030245126559262e-06,
|
| 2122 |
+
"loss": 0.4843,
|
| 2123 |
+
"step": 302
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 1.7380497131931167,
|
| 2127 |
+
"grad_norm": 0.402310432836796,
|
| 2128 |
+
"learning_rate": 8.963598178726254e-06,
|
| 2129 |
+
"loss": 0.3751,
|
| 2130 |
+
"step": 303
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 1.7437858508604207,
|
| 2134 |
+
"grad_norm": 0.438445094499713,
|
| 2135 |
+
"learning_rate": 8.896997733853904e-06,
|
| 2136 |
+
"loss": 0.4418,
|
| 2137 |
+
"step": 304
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 1.7495219885277247,
|
| 2141 |
+
"grad_norm": 0.440914495029627,
|
| 2142 |
+
"learning_rate": 8.830446780279175e-06,
|
| 2143 |
+
"loss": 0.4854,
|
| 2144 |
+
"step": 305
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 1.7552581261950286,
|
| 2148 |
+
"grad_norm": 0.38274381913523997,
|
| 2149 |
+
"learning_rate": 8.763948304118368e-06,
|
| 2150 |
+
"loss": 0.4132,
|
| 2151 |
+
"step": 306
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 1.7609942638623326,
|
| 2155 |
+
"grad_norm": 0.46352350104923684,
|
| 2156 |
+
"learning_rate": 8.697505289133146e-06,
|
| 2157 |
+
"loss": 0.4512,
|
| 2158 |
+
"step": 307
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 1.7667304015296366,
|
| 2162 |
+
"grad_norm": 0.3717919590029904,
|
| 2163 |
+
"learning_rate": 8.631120716596642e-06,
|
| 2164 |
+
"loss": 0.3694,
|
| 2165 |
+
"step": 308
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 1.7724665391969407,
|
| 2169 |
+
"grad_norm": 0.3770809977975791,
|
| 2170 |
+
"learning_rate": 8.564797565159715e-06,
|
| 2171 |
+
"loss": 0.4099,
|
| 2172 |
+
"step": 309
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 1.7782026768642447,
|
| 2176 |
+
"grad_norm": 0.430430532759824,
|
| 2177 |
+
"learning_rate": 8.498538810717267e-06,
|
| 2178 |
+
"loss": 0.4327,
|
| 2179 |
+
"step": 310
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 1.783938814531549,
|
| 2183 |
+
"grad_norm": 0.4097548018133223,
|
| 2184 |
+
"learning_rate": 8.432347426274739e-06,
|
| 2185 |
+
"loss": 0.4541,
|
| 2186 |
+
"step": 311
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 1.7896749521988529,
|
| 2190 |
+
"grad_norm": 0.4332198498570345,
|
| 2191 |
+
"learning_rate": 8.366226381814698e-06,
|
| 2192 |
+
"loss": 0.4592,
|
| 2193 |
+
"step": 312
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 1.7954110898661568,
|
| 2197 |
+
"grad_norm": 0.45602674402620935,
|
| 2198 |
+
"learning_rate": 8.300178644163593e-06,
|
| 2199 |
+
"loss": 0.4062,
|
| 2200 |
+
"step": 313
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 1.8011472275334608,
|
| 2204 |
+
"grad_norm": 0.4580406538791895,
|
| 2205 |
+
"learning_rate": 8.234207176858615e-06,
|
| 2206 |
+
"loss": 0.4568,
|
| 2207 |
+
"step": 314
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 1.8068833652007648,
|
| 2211 |
+
"grad_norm": 0.39852421924765624,
|
| 2212 |
+
"learning_rate": 8.168314940014742e-06,
|
| 2213 |
+
"loss": 0.4277,
|
| 2214 |
+
"step": 315
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 1.8126195028680687,
|
| 2218 |
+
"grad_norm": 0.4439622106678351,
|
| 2219 |
+
"learning_rate": 8.102504890191892e-06,
|
| 2220 |
+
"loss": 0.46,
|
| 2221 |
+
"step": 316
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 1.8183556405353727,
|
| 2225 |
+
"grad_norm": 0.3778756198759015,
|
| 2226 |
+
"learning_rate": 8.036779980262312e-06,
|
| 2227 |
+
"loss": 0.4223,
|
| 2228 |
+
"step": 317
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 1.8240917782026769,
|
| 2232 |
+
"grad_norm": 0.40171442544483804,
|
| 2233 |
+
"learning_rate": 7.971143159278025e-06,
|
| 2234 |
+
"loss": 0.4132,
|
| 2235 |
+
"step": 318
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 1.8298279158699808,
|
| 2239 |
+
"grad_norm": 0.41526173759534935,
|
| 2240 |
+
"learning_rate": 7.905597372338558e-06,
|
| 2241 |
+
"loss": 0.4391,
|
| 2242 |
+
"step": 319
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 1.835564053537285,
|
| 2246 |
+
"grad_norm": 0.38573779193525365,
|
| 2247 |
+
"learning_rate": 7.840145560458756e-06,
|
| 2248 |
+
"loss": 0.3864,
|
| 2249 |
+
"step": 320
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 1.841300191204589,
|
| 2253 |
+
"grad_norm": 0.41638087217030073,
|
| 2254 |
+
"learning_rate": 7.774790660436857e-06,
|
| 2255 |
+
"loss": 0.414,
|
| 2256 |
+
"step": 321
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 1.847036328871893,
|
| 2260 |
+
"grad_norm": 0.4526590955714858,
|
| 2261 |
+
"learning_rate": 7.709535604722685e-06,
|
| 2262 |
+
"loss": 0.4179,
|
| 2263 |
+
"step": 322
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 1.852772466539197,
|
| 2267 |
+
"grad_norm": 0.4080092443695909,
|
| 2268 |
+
"learning_rate": 7.644383321286095e-06,
|
| 2269 |
+
"loss": 0.4631,
|
| 2270 |
+
"step": 323
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 1.8585086042065009,
|
| 2274 |
+
"grad_norm": 0.38618254565446664,
|
| 2275 |
+
"learning_rate": 7.579336733485584e-06,
|
| 2276 |
+
"loss": 0.452,
|
| 2277 |
+
"step": 324
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 1.8642447418738048,
|
| 2281 |
+
"grad_norm": 0.374394051360769,
|
| 2282 |
+
"learning_rate": 7.514398759937135e-06,
|
| 2283 |
+
"loss": 0.4218,
|
| 2284 |
+
"step": 325
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 1.869980879541109,
|
| 2288 |
+
"grad_norm": 0.44912456810666246,
|
| 2289 |
+
"learning_rate": 7.449572314383237e-06,
|
| 2290 |
+
"loss": 0.4457,
|
| 2291 |
+
"step": 326
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 1.875717017208413,
|
| 2295 |
+
"grad_norm": 0.4076165446267555,
|
| 2296 |
+
"learning_rate": 7.384860305562171e-06,
|
| 2297 |
+
"loss": 0.4096,
|
| 2298 |
+
"step": 327
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 1.8814531548757172,
|
| 2302 |
+
"grad_norm": 0.41473117525412695,
|
| 2303 |
+
"learning_rate": 7.320265637077473e-06,
|
| 2304 |
+
"loss": 0.4833,
|
| 2305 |
+
"step": 328
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 1.8871892925430211,
|
| 2309 |
+
"grad_norm": 0.37006808150559667,
|
| 2310 |
+
"learning_rate": 7.255791207267679e-06,
|
| 2311 |
+
"loss": 0.3945,
|
| 2312 |
+
"step": 329
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 1.892925430210325,
|
| 2316 |
+
"grad_norm": 0.3956745154722134,
|
| 2317 |
+
"learning_rate": 7.191439909076243e-06,
|
| 2318 |
+
"loss": 0.4165,
|
| 2319 |
+
"step": 330
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 1.898661567877629,
|
| 2323 |
+
"grad_norm": 0.40315743805579995,
|
| 2324 |
+
"learning_rate": 7.127214629921765e-06,
|
| 2325 |
+
"loss": 0.408,
|
| 2326 |
+
"step": 331
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 1.904397705544933,
|
| 2330 |
+
"grad_norm": 0.40167879385408956,
|
| 2331 |
+
"learning_rate": 7.0631182515684e-06,
|
| 2332 |
+
"loss": 0.4069,
|
| 2333 |
+
"step": 332
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 1.910133843212237,
|
| 2337 |
+
"grad_norm": 0.4110824483433327,
|
| 2338 |
+
"learning_rate": 6.999153649996595e-06,
|
| 2339 |
+
"loss": 0.4033,
|
| 2340 |
+
"step": 333
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 1.915869980879541,
|
| 2344 |
+
"grad_norm": 0.4023551235142921,
|
| 2345 |
+
"learning_rate": 6.935323695274002e-06,
|
| 2346 |
+
"loss": 0.4244,
|
| 2347 |
+
"step": 334
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 1.9216061185468452,
|
| 2351 |
+
"grad_norm": 0.4076645867761283,
|
| 2352 |
+
"learning_rate": 6.871631251426729e-06,
|
| 2353 |
+
"loss": 0.4935,
|
| 2354 |
+
"step": 335
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 1.9273422562141491,
|
| 2358 |
+
"grad_norm": 0.4040181892237224,
|
| 2359 |
+
"learning_rate": 6.808079176310826e-06,
|
| 2360 |
+
"loss": 0.3796,
|
| 2361 |
+
"step": 336
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 1.9330783938814533,
|
| 2365 |
+
"grad_norm": 0.4188015305713764,
|
| 2366 |
+
"learning_rate": 6.744670321484044e-06,
|
| 2367 |
+
"loss": 0.406,
|
| 2368 |
+
"step": 337
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 1.9388145315487573,
|
| 2372 |
+
"grad_norm": 0.4143577555520623,
|
| 2373 |
+
"learning_rate": 6.681407532077895e-06,
|
| 2374 |
+
"loss": 0.4365,
|
| 2375 |
+
"step": 338
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 1.9445506692160612,
|
| 2379 |
+
"grad_norm": 0.3764270258745421,
|
| 2380 |
+
"learning_rate": 6.618293646669987e-06,
|
| 2381 |
+
"loss": 0.4366,
|
| 2382 |
+
"step": 339
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 1.9502868068833652,
|
| 2386 |
+
"grad_norm": 0.4210114400843068,
|
| 2387 |
+
"learning_rate": 6.555331497156671e-06,
|
| 2388 |
+
"loss": 0.4551,
|
| 2389 |
+
"step": 340
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 1.9560229445506692,
|
| 2393 |
+
"grad_norm": 0.4453880959527477,
|
| 2394 |
+
"learning_rate": 6.492523908625958e-06,
|
| 2395 |
+
"loss": 0.4564,
|
| 2396 |
+
"step": 341
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 1.9617590822179731,
|
| 2400 |
+
"grad_norm": 0.4037182452331645,
|
| 2401 |
+
"learning_rate": 6.429873699230771e-06,
|
| 2402 |
+
"loss": 0.4411,
|
| 2403 |
+
"step": 342
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 1.967495219885277,
|
| 2407 |
+
"grad_norm": 0.3808388760596802,
|
| 2408 |
+
"learning_rate": 6.3673836800624775e-06,
|
| 2409 |
+
"loss": 0.3952,
|
| 2410 |
+
"step": 343
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 1.9732313575525813,
|
| 2414 |
+
"grad_norm": 0.41924994933550513,
|
| 2415 |
+
"learning_rate": 6.30505665502479e-06,
|
| 2416 |
+
"loss": 0.3802,
|
| 2417 |
+
"step": 344
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 1.9789674952198852,
|
| 2421 |
+
"grad_norm": 0.44573552239332564,
|
| 2422 |
+
"learning_rate": 6.242895420707917e-06,
|
| 2423 |
+
"loss": 0.4741,
|
| 2424 |
+
"step": 345
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 1.9847036328871894,
|
| 2428 |
+
"grad_norm": 0.44311634814108447,
|
| 2429 |
+
"learning_rate": 6.180902766263113e-06,
|
| 2430 |
+
"loss": 0.4568,
|
| 2431 |
+
"step": 346
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 1.9904397705544934,
|
| 2435 |
+
"grad_norm": 0.403141283403809,
|
| 2436 |
+
"learning_rate": 6.119081473277502e-06,
|
| 2437 |
+
"loss": 0.4439,
|
| 2438 |
+
"step": 347
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 1.9961759082217974,
|
| 2442 |
+
"grad_norm": 0.3696751925678653,
|
| 2443 |
+
"learning_rate": 6.057434315649304e-06,
|
| 2444 |
+
"loss": 0.4076,
|
| 2445 |
+
"step": 348
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 2.0019120458891013,
|
| 2449 |
+
"grad_norm": 0.8972308979080702,
|
| 2450 |
+
"learning_rate": 5.9959640594633304e-06,
|
| 2451 |
+
"loss": 0.6431,
|
| 2452 |
+
"step": 349
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 2.0076481835564053,
|
| 2456 |
+
"grad_norm": 0.4952120180229906,
|
| 2457 |
+
"learning_rate": 5.934673462866907e-06,
|
| 2458 |
+
"loss": 0.4089,
|
| 2459 |
+
"step": 350
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 2.0133843212237093,
|
| 2463 |
+
"grad_norm": 0.5208407249986332,
|
| 2464 |
+
"learning_rate": 5.873565275946088e-06,
|
| 2465 |
+
"loss": 0.3998,
|
| 2466 |
+
"step": 351
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 2.019120458891013,
|
| 2470 |
+
"grad_norm": 0.5507106804898475,
|
| 2471 |
+
"learning_rate": 5.812642240602289e-06,
|
| 2472 |
+
"loss": 0.3835,
|
| 2473 |
+
"step": 352
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 2.024856596558317,
|
| 2477 |
+
"grad_norm": 0.5141610472881658,
|
| 2478 |
+
"learning_rate": 5.7519070904292255e-06,
|
| 2479 |
+
"loss": 0.3859,
|
| 2480 |
+
"step": 353
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 2.0305927342256216,
|
| 2484 |
+
"grad_norm": 0.46442820428267456,
|
| 2485 |
+
"learning_rate": 5.6913625505902966e-06,
|
| 2486 |
+
"loss": 0.364,
|
| 2487 |
+
"step": 354
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 2.0363288718929256,
|
| 2491 |
+
"grad_norm": 0.4645855145414816,
|
| 2492 |
+
"learning_rate": 5.631011337696272e-06,
|
| 2493 |
+
"loss": 0.3577,
|
| 2494 |
+
"step": 355
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 2.0420650095602295,
|
| 2498 |
+
"grad_norm": 0.5129160739607502,
|
| 2499 |
+
"learning_rate": 5.570856159683418e-06,
|
| 2500 |
+
"loss": 0.3708,
|
| 2501 |
+
"step": 356
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 2.0478011472275335,
|
| 2505 |
+
"grad_norm": 0.44092823234948325,
|
| 2506 |
+
"learning_rate": 5.510899715691984e-06,
|
| 2507 |
+
"loss": 0.3954,
|
| 2508 |
+
"step": 357
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 2.0535372848948374,
|
| 2512 |
+
"grad_norm": 0.4307147154928634,
|
| 2513 |
+
"learning_rate": 5.451144695945116e-06,
|
| 2514 |
+
"loss": 0.3767,
|
| 2515 |
+
"step": 358
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 2.0592734225621414,
|
| 2519 |
+
"grad_norm": 0.3954871188814476,
|
| 2520 |
+
"learning_rate": 5.391593781628109e-06,
|
| 2521 |
+
"loss": 0.3096,
|
| 2522 |
+
"step": 359
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 2.0650095602294454,
|
| 2526 |
+
"grad_norm": 0.45723009765562855,
|
| 2527 |
+
"learning_rate": 5.332249644768142e-06,
|
| 2528 |
+
"loss": 0.3795,
|
| 2529 |
+
"step": 360
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 2.0707456978967493,
|
| 2533 |
+
"grad_norm": 0.41297055229228974,
|
| 2534 |
+
"learning_rate": 5.273114948114346e-06,
|
| 2535 |
+
"loss": 0.3177,
|
| 2536 |
+
"step": 361
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 2.0764818355640537,
|
| 2540 |
+
"grad_norm": 0.4106872355240582,
|
| 2541 |
+
"learning_rate": 5.214192345018374e-06,
|
| 2542 |
+
"loss": 0.3515,
|
| 2543 |
+
"step": 362
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 2.0822179732313577,
|
| 2547 |
+
"grad_norm": 0.42883512067861784,
|
| 2548 |
+
"learning_rate": 5.1554844793153005e-06,
|
| 2549 |
+
"loss": 0.4205,
|
| 2550 |
+
"step": 363
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 2.0879541108986617,
|
| 2554 |
+
"grad_norm": 0.3815636168034508,
|
| 2555 |
+
"learning_rate": 5.096993985205023e-06,
|
| 2556 |
+
"loss": 0.3495,
|
| 2557 |
+
"step": 364
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 2.0936902485659656,
|
| 2561 |
+
"grad_norm": 0.40415485231377263,
|
| 2562 |
+
"learning_rate": 5.038723487134049e-06,
|
| 2563 |
+
"loss": 0.3969,
|
| 2564 |
+
"step": 365
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 2.0994263862332696,
|
| 2568 |
+
"grad_norm": 0.3948215113870869,
|
| 2569 |
+
"learning_rate": 4.9806755996777565e-06,
|
| 2570 |
+
"loss": 0.37,
|
| 2571 |
+
"step": 366
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 2.1051625239005736,
|
| 2575 |
+
"grad_norm": 0.3961343148482674,
|
| 2576 |
+
"learning_rate": 4.92285292742307e-06,
|
| 2577 |
+
"loss": 0.4183,
|
| 2578 |
+
"step": 367
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 2.1108986615678775,
|
| 2582 |
+
"grad_norm": 0.35475813700193604,
|
| 2583 |
+
"learning_rate": 4.865258064851579e-06,
|
| 2584 |
+
"loss": 0.3405,
|
| 2585 |
+
"step": 368
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 2.1166347992351815,
|
| 2589 |
+
"grad_norm": 0.38458880165760984,
|
| 2590 |
+
"learning_rate": 4.807893596223152e-06,
|
| 2591 |
+
"loss": 0.3702,
|
| 2592 |
+
"step": 369
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 2.1223709369024855,
|
| 2596 |
+
"grad_norm": 0.4018427850015987,
|
| 2597 |
+
"learning_rate": 4.75076209545996e-06,
|
| 2598 |
+
"loss": 0.3882,
|
| 2599 |
+
"step": 370
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 2.12810707456979,
|
| 2603 |
+
"grad_norm": 0.3649236891798529,
|
| 2604 |
+
"learning_rate": 4.693866126030995e-06,
|
| 2605 |
+
"loss": 0.3192,
|
| 2606 |
+
"step": 371
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 2.133843212237094,
|
| 2610 |
+
"grad_norm": 0.4081333849276713,
|
| 2611 |
+
"learning_rate": 4.637208240837042e-06,
|
| 2612 |
+
"loss": 0.3954,
|
| 2613 |
+
"step": 372
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 2.139579349904398,
|
| 2617 |
+
"grad_norm": 0.35987845861598494,
|
| 2618 |
+
"learning_rate": 4.580790982096149e-06,
|
| 2619 |
+
"loss": 0.3096,
|
| 2620 |
+
"step": 373
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 2.1453154875717018,
|
| 2624 |
+
"grad_norm": 0.430627026676002,
|
| 2625 |
+
"learning_rate": 4.5246168812295286e-06,
|
| 2626 |
+
"loss": 0.4717,
|
| 2627 |
+
"step": 374
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 2.1510516252390057,
|
| 2631 |
+
"grad_norm": 0.3338158346155251,
|
| 2632 |
+
"learning_rate": 4.468688458748006e-06,
|
| 2633 |
+
"loss": 0.2981,
|
| 2634 |
+
"step": 375
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 2.1567877629063097,
|
| 2638 |
+
"grad_norm": 0.38106855044885324,
|
| 2639 |
+
"learning_rate": 4.4130082241388974e-06,
|
| 2640 |
+
"loss": 0.3767,
|
| 2641 |
+
"step": 376
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 2.1625239005736137,
|
| 2645 |
+
"grad_norm": 0.36059174781408515,
|
| 2646 |
+
"learning_rate": 4.357578675753432e-06,
|
| 2647 |
+
"loss": 0.3631,
|
| 2648 |
+
"step": 377
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 2.1682600382409176,
|
| 2652 |
+
"grad_norm": 0.3882154722055546,
|
| 2653 |
+
"learning_rate": 4.302402300694636e-06,
|
| 2654 |
+
"loss": 0.3738,
|
| 2655 |
+
"step": 378
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 2.173996175908222,
|
| 2659 |
+
"grad_norm": 0.3978520905363434,
|
| 2660 |
+
"learning_rate": 4.247481574705744e-06,
|
| 2661 |
+
"loss": 0.3309,
|
| 2662 |
+
"step": 379
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 2.179732313575526,
|
| 2666 |
+
"grad_norm": 0.3930085794069971,
|
| 2667 |
+
"learning_rate": 4.192818962059112e-06,
|
| 2668 |
+
"loss": 0.3395,
|
| 2669 |
+
"step": 380
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 2.18546845124283,
|
| 2673 |
+
"grad_norm": 0.3679820549907332,
|
| 2674 |
+
"learning_rate": 4.138416915445656e-06,
|
| 2675 |
+
"loss": 0.3521,
|
| 2676 |
+
"step": 381
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 2.191204588910134,
|
| 2680 |
+
"grad_norm": 0.3851345683393216,
|
| 2681 |
+
"learning_rate": 4.084277875864776e-06,
|
| 2682 |
+
"loss": 0.379,
|
| 2683 |
+
"step": 382
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 2.196940726577438,
|
| 2687 |
+
"grad_norm": 0.44781870089318787,
|
| 2688 |
+
"learning_rate": 4.030404272514864e-06,
|
| 2689 |
+
"loss": 0.3803,
|
| 2690 |
+
"step": 383
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 2.202676864244742,
|
| 2694 |
+
"grad_norm": 0.39666830508510403,
|
| 2695 |
+
"learning_rate": 3.97679852268427e-06,
|
| 2696 |
+
"loss": 0.3622,
|
| 2697 |
+
"step": 384
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 2.208413001912046,
|
| 2701 |
+
"grad_norm": 0.3976428991458035,
|
| 2702 |
+
"learning_rate": 3.923463031642873e-06,
|
| 2703 |
+
"loss": 0.3673,
|
| 2704 |
+
"step": 385
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 2.21414913957935,
|
| 2708 |
+
"grad_norm": 0.34487643216663577,
|
| 2709 |
+
"learning_rate": 3.870400192534128e-06,
|
| 2710 |
+
"loss": 0.3647,
|
| 2711 |
+
"step": 386
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 2.2198852772466537,
|
| 2715 |
+
"grad_norm": 0.37349012951330834,
|
| 2716 |
+
"learning_rate": 3.81761238626771e-06,
|
| 2717 |
+
"loss": 0.3822,
|
| 2718 |
+
"step": 387
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 2.2256214149139577,
|
| 2722 |
+
"grad_norm": 0.40320901857132385,
|
| 2723 |
+
"learning_rate": 3.7651019814126656e-06,
|
| 2724 |
+
"loss": 0.4045,
|
| 2725 |
+
"step": 388
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 2.231357552581262,
|
| 2729 |
+
"grad_norm": 0.38833879539410954,
|
| 2730 |
+
"learning_rate": 3.712871334091154e-06,
|
| 2731 |
+
"loss": 0.3614,
|
| 2732 |
+
"step": 389
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 2.237093690248566,
|
| 2736 |
+
"grad_norm": 0.37829756392287156,
|
| 2737 |
+
"learning_rate": 3.6609227878727062e-06,
|
| 2738 |
+
"loss": 0.3271,
|
| 2739 |
+
"step": 390
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 2.24282982791587,
|
| 2743 |
+
"grad_norm": 0.3579239446318294,
|
| 2744 |
+
"learning_rate": 3.609258673669097e-06,
|
| 2745 |
+
"loss": 0.3428,
|
| 2746 |
+
"step": 391
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 2.248565965583174,
|
| 2750 |
+
"grad_norm": 0.3686157523103381,
|
| 2751 |
+
"learning_rate": 3.5578813096297293e-06,
|
| 2752 |
+
"loss": 0.3861,
|
| 2753 |
+
"step": 392
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 2.254302103250478,
|
| 2757 |
+
"grad_norm": 0.4030570129715361,
|
| 2758 |
+
"learning_rate": 3.5067930010376484e-06,
|
| 2759 |
+
"loss": 0.42,
|
| 2760 |
+
"step": 393
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 2.260038240917782,
|
| 2764 |
+
"grad_norm": 0.37486030062507975,
|
| 2765 |
+
"learning_rate": 3.4559960402060764e-06,
|
| 2766 |
+
"loss": 0.3634,
|
| 2767 |
+
"step": 394
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 2.265774378585086,
|
| 2771 |
+
"grad_norm": 0.398111741855188,
|
| 2772 |
+
"learning_rate": 3.4054927063755793e-06,
|
| 2773 |
+
"loss": 0.4019,
|
| 2774 |
+
"step": 395
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 2.27151051625239,
|
| 2778 |
+
"grad_norm": 0.3742173600629292,
|
| 2779 |
+
"learning_rate": 3.355285265611784e-06,
|
| 2780 |
+
"loss": 0.3547,
|
| 2781 |
+
"step": 396
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 2.2772466539196943,
|
| 2785 |
+
"grad_norm": 0.4098405397196824,
|
| 2786 |
+
"learning_rate": 3.3053759707037116e-06,
|
| 2787 |
+
"loss": 0.3624,
|
| 2788 |
+
"step": 397
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 2.2829827915869982,
|
| 2792 |
+
"grad_norm": 0.42987789489488126,
|
| 2793 |
+
"learning_rate": 3.2557670610626924e-06,
|
| 2794 |
+
"loss": 0.3771,
|
| 2795 |
+
"step": 398
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 2.288718929254302,
|
| 2799 |
+
"grad_norm": 0.37950071404433056,
|
| 2800 |
+
"learning_rate": 3.2064607626218737e-06,
|
| 2801 |
+
"loss": 0.3793,
|
| 2802 |
+
"step": 399
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 2.294455066921606,
|
| 2806 |
+
"grad_norm": 0.38352933857672544,
|
| 2807 |
+
"learning_rate": 3.157459287736362e-06,
|
| 2808 |
+
"loss": 0.3784,
|
| 2809 |
+
"step": 400
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 2.30019120458891,
|
| 2813 |
+
"grad_norm": 0.3915772618856507,
|
| 2814 |
+
"learning_rate": 3.1087648350839382e-06,
|
| 2815 |
+
"loss": 0.3892,
|
| 2816 |
+
"step": 401
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 2.305927342256214,
|
| 2820 |
+
"grad_norm": 0.38035400732236047,
|
| 2821 |
+
"learning_rate": 3.0603795895664125e-06,
|
| 2822 |
+
"loss": 0.3596,
|
| 2823 |
+
"step": 402
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 2.311663479923518,
|
| 2827 |
+
"grad_norm": 0.3966721118627521,
|
| 2828 |
+
"learning_rate": 3.0123057222115835e-06,
|
| 2829 |
+
"loss": 0.3933,
|
| 2830 |
+
"step": 403
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 2.317399617590822,
|
| 2834 |
+
"grad_norm": 0.3928138522564511,
|
| 2835 |
+
"learning_rate": 2.9645453900758415e-06,
|
| 2836 |
+
"loss": 0.3858,
|
| 2837 |
+
"step": 404
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 2.323135755258126,
|
| 2841 |
+
"grad_norm": 0.3574946707024923,
|
| 2842 |
+
"learning_rate": 2.9171007361473512e-06,
|
| 2843 |
+
"loss": 0.3503,
|
| 2844 |
+
"step": 405
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 2.3288718929254304,
|
| 2848 |
+
"grad_norm": 0.3638977416533278,
|
| 2849 |
+
"learning_rate": 2.869973889249933e-06,
|
| 2850 |
+
"loss": 0.3211,
|
| 2851 |
+
"step": 406
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 2.3346080305927344,
|
| 2855 |
+
"grad_norm": 0.396313353596241,
|
| 2856 |
+
"learning_rate": 2.8231669639475068e-06,
|
| 2857 |
+
"loss": 0.3283,
|
| 2858 |
+
"step": 407
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 2.3403441682600383,
|
| 2862 |
+
"grad_norm": 0.4350084113143233,
|
| 2863 |
+
"learning_rate": 2.7766820604492474e-06,
|
| 2864 |
+
"loss": 0.3694,
|
| 2865 |
+
"step": 408
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 2.3460803059273423,
|
| 2869 |
+
"grad_norm": 0.36417177098203485,
|
| 2870 |
+
"learning_rate": 2.7305212645153213e-06,
|
| 2871 |
+
"loss": 0.3277,
|
| 2872 |
+
"step": 409
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 2.3518164435946463,
|
| 2876 |
+
"grad_norm": 0.36228834197930426,
|
| 2877 |
+
"learning_rate": 2.6846866473633126e-06,
|
| 2878 |
+
"loss": 0.3648,
|
| 2879 |
+
"step": 410
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 2.35755258126195,
|
| 2883 |
+
"grad_norm": 0.37217108995589493,
|
| 2884 |
+
"learning_rate": 2.6391802655752853e-06,
|
| 2885 |
+
"loss": 0.3553,
|
| 2886 |
+
"step": 411
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 2.363288718929254,
|
| 2890 |
+
"grad_norm": 0.3796921652078883,
|
| 2891 |
+
"learning_rate": 2.594004161005511e-06,
|
| 2892 |
+
"loss": 0.3953,
|
| 2893 |
+
"step": 412
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 2.369024856596558,
|
| 2897 |
+
"grad_norm": 0.37234421223477726,
|
| 2898 |
+
"learning_rate": 2.5491603606888384e-06,
|
| 2899 |
+
"loss": 0.3431,
|
| 2900 |
+
"step": 413
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 2.3747609942638626,
|
| 2904 |
+
"grad_norm": 0.3479981334692919,
|
| 2905 |
+
"learning_rate": 2.50465087674976e-06,
|
| 2906 |
+
"loss": 0.3587,
|
| 2907 |
+
"step": 414
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 2.3804971319311665,
|
| 2911 |
+
"grad_norm": 0.37874728957661286,
|
| 2912 |
+
"learning_rate": 2.460477706312103e-06,
|
| 2913 |
+
"loss": 0.3871,
|
| 2914 |
+
"step": 415
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 2.3862332695984705,
|
| 2918 |
+
"grad_norm": 0.3570282223249849,
|
| 2919 |
+
"learning_rate": 2.4166428314094514e-06,
|
| 2920 |
+
"loss": 0.3796,
|
| 2921 |
+
"step": 416
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 2.3919694072657744,
|
| 2925 |
+
"grad_norm": 0.3547259024756538,
|
| 2926 |
+
"learning_rate": 2.373148218896182e-06,
|
| 2927 |
+
"loss": 0.3723,
|
| 2928 |
+
"step": 417
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 2.3977055449330784,
|
| 2932 |
+
"grad_norm": 0.36243719147680803,
|
| 2933 |
+
"learning_rate": 2.32999582035923e-06,
|
| 2934 |
+
"loss": 0.375,
|
| 2935 |
+
"step": 418
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 2.4034416826003824,
|
| 2939 |
+
"grad_norm": 0.3397982054868222,
|
| 2940 |
+
"learning_rate": 2.2871875720305158e-06,
|
| 2941 |
+
"loss": 0.359,
|
| 2942 |
+
"step": 419
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 2.4091778202676863,
|
| 2946 |
+
"grad_norm": 0.3706667244009237,
|
| 2947 |
+
"learning_rate": 2.244725394700079e-06,
|
| 2948 |
+
"loss": 0.4019,
|
| 2949 |
+
"step": 420
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 2.4149139579349903,
|
| 2953 |
+
"grad_norm": 0.3521855735970329,
|
| 2954 |
+
"learning_rate": 2.2026111936298687e-06,
|
| 2955 |
+
"loss": 0.339,
|
| 2956 |
+
"step": 421
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 2.4206500956022943,
|
| 2960 |
+
"grad_norm": 0.36197027269129106,
|
| 2961 |
+
"learning_rate": 2.160846858468285e-06,
|
| 2962 |
+
"loss": 0.4068,
|
| 2963 |
+
"step": 422
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 2.4263862332695982,
|
| 2967 |
+
"grad_norm": 0.3486550600296997,
|
| 2968 |
+
"learning_rate": 2.119434263165361e-06,
|
| 2969 |
+
"loss": 0.3472,
|
| 2970 |
+
"step": 423
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 2.4321223709369026,
|
| 2974 |
+
"grad_norm": 0.36592179113764045,
|
| 2975 |
+
"learning_rate": 2.078375265888707e-06,
|
| 2976 |
+
"loss": 0.341,
|
| 2977 |
+
"step": 424
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 2.4378585086042066,
|
| 2981 |
+
"grad_norm": 0.381502441618795,
|
| 2982 |
+
"learning_rate": 2.0376717089401166e-06,
|
| 2983 |
+
"loss": 0.3826,
|
| 2984 |
+
"step": 425
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 2.4435946462715106,
|
| 2988 |
+
"grad_norm": 0.3690496531978919,
|
| 2989 |
+
"learning_rate": 1.9973254186729084e-06,
|
| 2990 |
+
"loss": 0.3562,
|
| 2991 |
+
"step": 426
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 2.4493307839388145,
|
| 2995 |
+
"grad_norm": 0.37416288175512286,
|
| 2996 |
+
"learning_rate": 1.9573382054099786e-06,
|
| 2997 |
+
"loss": 0.4103,
|
| 2998 |
+
"step": 427
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 2.4550669216061185,
|
| 3002 |
+
"grad_norm": 0.35395175975410714,
|
| 3003 |
+
"learning_rate": 1.917711863362581e-06,
|
| 3004 |
+
"loss": 0.3632,
|
| 3005 |
+
"step": 428
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 2.4608030592734225,
|
| 3009 |
+
"grad_norm": 0.36581947179840435,
|
| 3010 |
+
"learning_rate": 1.8784481705498014e-06,
|
| 3011 |
+
"loss": 0.4023,
|
| 3012 |
+
"step": 429
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 2.4665391969407264,
|
| 3016 |
+
"grad_norm": 0.33261262472276076,
|
| 3017 |
+
"learning_rate": 1.8395488887188007e-06,
|
| 3018 |
+
"loss": 0.3394,
|
| 3019 |
+
"step": 430
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 2.472275334608031,
|
| 3023 |
+
"grad_norm": 0.3731396992459488,
|
| 3024 |
+
"learning_rate": 1.8010157632657544e-06,
|
| 3025 |
+
"loss": 0.3784,
|
| 3026 |
+
"step": 431
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 2.478011472275335,
|
| 3030 |
+
"grad_norm": 0.33337120708056256,
|
| 3031 |
+
"learning_rate": 1.7628505231575321e-06,
|
| 3032 |
+
"loss": 0.372,
|
| 3033 |
+
"step": 432
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 2.4837476099426388,
|
| 3037 |
+
"grad_norm": 0.3638826195090011,
|
| 3038 |
+
"learning_rate": 1.7250548808541324e-06,
|
| 3039 |
+
"loss": 0.4246,
|
| 3040 |
+
"step": 433
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 2.4894837476099427,
|
| 3044 |
+
"grad_norm": 0.36142571281527636,
|
| 3045 |
+
"learning_rate": 1.687630532231833e-06,
|
| 3046 |
+
"loss": 0.3674,
|
| 3047 |
+
"step": 434
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 2.4952198852772467,
|
| 3051 |
+
"grad_norm": 0.41140218172654,
|
| 3052 |
+
"learning_rate": 1.6505791565071139e-06,
|
| 3053 |
+
"loss": 0.3625,
|
| 3054 |
+
"step": 435
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 2.5009560229445507,
|
| 3058 |
+
"grad_norm": 0.3451954148555015,
|
| 3059 |
+
"learning_rate": 1.6139024161612882e-06,
|
| 3060 |
+
"loss": 0.3604,
|
| 3061 |
+
"step": 436
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 2.5066921606118546,
|
| 3065 |
+
"grad_norm": 0.32260290363728555,
|
| 3066 |
+
"learning_rate": 1.577601956865934e-06,
|
| 3067 |
+
"loss": 0.3679,
|
| 3068 |
+
"step": 437
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 2.5124282982791586,
|
| 3072 |
+
"grad_norm": 0.34462182092709037,
|
| 3073 |
+
"learning_rate": 1.5416794074090258e-06,
|
| 3074 |
+
"loss": 0.3741,
|
| 3075 |
+
"step": 438
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 2.5181644359464626,
|
| 3079 |
+
"grad_norm": 0.4730909294396266,
|
| 3080 |
+
"learning_rate": 1.5061363796218787e-06,
|
| 3081 |
+
"loss": 0.3667,
|
| 3082 |
+
"step": 439
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 2.5239005736137665,
|
| 3086 |
+
"grad_norm": 0.3857798475766756,
|
| 3087 |
+
"learning_rate": 1.470974468306804e-06,
|
| 3088 |
+
"loss": 0.4242,
|
| 3089 |
+
"step": 440
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 2.5296367112810705,
|
| 3093 |
+
"grad_norm": 0.3620511836720628,
|
| 3094 |
+
"learning_rate": 1.4361952511655618e-06,
|
| 3095 |
+
"loss": 0.3954,
|
| 3096 |
+
"step": 441
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 2.535372848948375,
|
| 3100 |
+
"grad_norm": 0.330256905865385,
|
| 3101 |
+
"learning_rate": 1.4018002887285687e-06,
|
| 3102 |
+
"loss": 0.3261,
|
| 3103 |
+
"step": 442
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 2.541108986615679,
|
| 3107 |
+
"grad_norm": 0.3714557217698261,
|
| 3108 |
+
"learning_rate": 1.3677911242848807e-06,
|
| 3109 |
+
"loss": 0.3538,
|
| 3110 |
+
"step": 443
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 2.546845124282983,
|
| 3114 |
+
"grad_norm": 0.35180843150904273,
|
| 3115 |
+
"learning_rate": 1.334169283812936e-06,
|
| 3116 |
+
"loss": 0.3534,
|
| 3117 |
+
"step": 444
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 2.552581261950287,
|
| 3121 |
+
"grad_norm": 0.3461429617840913,
|
| 3122 |
+
"learning_rate": 1.300936275912098e-06,
|
| 3123 |
+
"loss": 0.3934,
|
| 3124 |
+
"step": 445
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 2.5583173996175907,
|
| 3128 |
+
"grad_norm": 0.3481927841057214,
|
| 3129 |
+
"learning_rate": 1.2680935917349524e-06,
|
| 3130 |
+
"loss": 0.3627,
|
| 3131 |
+
"step": 446
|
| 3132 |
+
},
|
| 3133 |
+
{
|
| 3134 |
+
"epoch": 2.5640535372848947,
|
| 3135 |
+
"grad_norm": 0.323481747190712,
|
| 3136 |
+
"learning_rate": 1.2356427049204122e-06,
|
| 3137 |
+
"loss": 0.3545,
|
| 3138 |
+
"step": 447
|
| 3139 |
+
},
|
| 3140 |
+
{
|
| 3141 |
+
"epoch": 2.569789674952199,
|
| 3142 |
+
"grad_norm": 0.3437913936704319,
|
| 3143 |
+
"learning_rate": 1.2035850715275865e-06,
|
| 3144 |
+
"loss": 0.3776,
|
| 3145 |
+
"step": 448
|
| 3146 |
+
},
|
| 3147 |
+
{
|
| 3148 |
+
"epoch": 2.575525812619503,
|
| 3149 |
+
"grad_norm": 0.3480873955124251,
|
| 3150 |
+
"learning_rate": 1.1719221299704497e-06,
|
| 3151 |
+
"loss": 0.3266,
|
| 3152 |
+
"step": 449
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 2.581261950286807,
|
| 3156 |
+
"grad_norm": 0.35378460832555164,
|
| 3157 |
+
"learning_rate": 1.1406553009533028e-06,
|
| 3158 |
+
"loss": 0.3243,
|
| 3159 |
+
"step": 450
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 2.586998087954111,
|
| 3163 |
+
"grad_norm": 0.3763656152527704,
|
| 3164 |
+
"learning_rate": 1.1097859874070294e-06,
|
| 3165 |
+
"loss": 0.4062,
|
| 3166 |
+
"step": 451
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 2.592734225621415,
|
| 3170 |
+
"grad_norm": 0.3569062431922278,
|
| 3171 |
+
"learning_rate": 1.0793155744261352e-06,
|
| 3172 |
+
"loss": 0.3596,
|
| 3173 |
+
"step": 452
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 2.598470363288719,
|
| 3177 |
+
"grad_norm": 0.3428907211304672,
|
| 3178 |
+
"learning_rate": 1.0492454292066178e-06,
|
| 3179 |
+
"loss": 0.3834,
|
| 3180 |
+
"step": 453
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 2.604206500956023,
|
| 3184 |
+
"grad_norm": 0.33957138193960923,
|
| 3185 |
+
"learning_rate": 1.0195769009845992e-06,
|
| 3186 |
+
"loss": 0.3485,
|
| 3187 |
+
"step": 454
|
| 3188 |
+
},
|
| 3189 |
+
{
|
| 3190 |
+
"epoch": 2.609942638623327,
|
| 3191 |
+
"grad_norm": 0.3506861012073475,
|
| 3192 |
+
"learning_rate": 9.903113209758098e-07,
|
| 3193 |
+
"loss": 0.401,
|
| 3194 |
+
"step": 455
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 2.615678776290631,
|
| 3198 |
+
"grad_norm": 0.35438156085082345,
|
| 3199 |
+
"learning_rate": 9.614500023158335e-07,
|
| 3200 |
+
"loss": 0.3622,
|
| 3201 |
+
"step": 456
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 2.621414913957935,
|
| 3205 |
+
"grad_norm": 0.3235710698929606,
|
| 3206 |
+
"learning_rate": 9.329942400012059e-07,
|
| 3207 |
+
"loss": 0.3383,
|
| 3208 |
+
"step": 457
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 2.6271510516252388,
|
| 3212 |
+
"grad_norm": 0.32641057302722415,
|
| 3213 |
+
"learning_rate": 9.049453108312967e-07,
|
| 3214 |
+
"loss": 0.3342,
|
| 3215 |
+
"step": 458
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 2.632887189292543,
|
| 3219 |
+
"grad_norm": 0.3738134966287739,
|
| 3220 |
+
"learning_rate": 8.773044733510338e-07,
|
| 3221 |
+
"loss": 0.3916,
|
| 3222 |
+
"step": 459
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 2.638623326959847,
|
| 3226 |
+
"grad_norm": 0.37276281277800666,
|
| 3227 |
+
"learning_rate": 8.50072967794413e-07,
|
| 3228 |
+
"loss": 0.398,
|
| 3229 |
+
"step": 460
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 2.644359464627151,
|
| 3233 |
+
"grad_norm": 0.35059967338877185,
|
| 3234 |
+
"learning_rate": 8.232520160288704e-07,
|
| 3235 |
+
"loss": 0.3666,
|
| 3236 |
+
"step": 461
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 2.650095602294455,
|
| 3240 |
+
"grad_norm": 0.34073212808078274,
|
| 3241 |
+
"learning_rate": 7.96842821500442e-07,
|
| 3242 |
+
"loss": 0.3533,
|
| 3243 |
+
"step": 462
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 2.655831739961759,
|
| 3247 |
+
"grad_norm": 0.3648080091480956,
|
| 3248 |
+
"learning_rate": 7.708465691797718e-07,
|
| 3249 |
+
"loss": 0.3904,
|
| 3250 |
+
"step": 463
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 2.661567877629063,
|
| 3254 |
+
"grad_norm": 0.3662958371430221,
|
| 3255 |
+
"learning_rate": 7.452644255089425e-07,
|
| 3256 |
+
"loss": 0.366,
|
| 3257 |
+
"step": 464
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 2.667304015296367,
|
| 3261 |
+
"grad_norm": 0.33496912547986984,
|
| 3262 |
+
"learning_rate": 7.20097538349136e-07,
|
| 3263 |
+
"loss": 0.3518,
|
| 3264 |
+
"step": 465
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 2.6730401529636714,
|
| 3268 |
+
"grad_norm": 0.3447271042188862,
|
| 3269 |
+
"learning_rate": 6.953470369291349e-07,
|
| 3270 |
+
"loss": 0.3531,
|
| 3271 |
+
"step": 466
|
| 3272 |
+
},
|
| 3273 |
+
{
|
| 3274 |
+
"epoch": 2.6787762906309753,
|
| 3275 |
+
"grad_norm": 0.3460068944711527,
|
| 3276 |
+
"learning_rate": 6.710140317946424e-07,
|
| 3277 |
+
"loss": 0.3756,
|
| 3278 |
+
"step": 467
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 2.6845124282982793,
|
| 3282 |
+
"grad_norm": 0.34479127909993756,
|
| 3283 |
+
"learning_rate": 6.470996147584684e-07,
|
| 3284 |
+
"loss": 0.3952,
|
| 3285 |
+
"step": 468
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 2.6902485659655833,
|
| 3289 |
+
"grad_norm": 0.3555895183537466,
|
| 3290 |
+
"learning_rate": 6.236048588515242e-07,
|
| 3291 |
+
"loss": 0.3534,
|
| 3292 |
+
"step": 469
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 2.6959847036328872,
|
| 3296 |
+
"grad_norm": 0.36902112669931375,
|
| 3297 |
+
"learning_rate": 6.005308182746906e-07,
|
| 3298 |
+
"loss": 0.3591,
|
| 3299 |
+
"step": 470
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 2.701720841300191,
|
| 3303 |
+
"grad_norm": 0.35072958260638964,
|
| 3304 |
+
"learning_rate": 5.778785283515054e-07,
|
| 3305 |
+
"loss": 0.3582,
|
| 3306 |
+
"step": 471
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 2.707456978967495,
|
| 3310 |
+
"grad_norm": 0.3435970056521967,
|
| 3311 |
+
"learning_rate": 5.556490054817132e-07,
|
| 3312 |
+
"loss": 0.3988,
|
| 3313 |
+
"step": 472
|
| 3314 |
+
},
|
| 3315 |
+
{
|
| 3316 |
+
"epoch": 2.713193116634799,
|
| 3317 |
+
"grad_norm": 0.3360949092846702,
|
| 3318 |
+
"learning_rate": 5.33843247095659e-07,
|
| 3319 |
+
"loss": 0.3857,
|
| 3320 |
+
"step": 473
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 2.718929254302103,
|
| 3324 |
+
"grad_norm": 0.36687321532000483,
|
| 3325 |
+
"learning_rate": 5.124622316095384e-07,
|
| 3326 |
+
"loss": 0.398,
|
| 3327 |
+
"step": 474
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 2.724665391969407,
|
| 3331 |
+
"grad_norm": 0.34930058861444474,
|
| 3332 |
+
"learning_rate": 4.91506918381488e-07,
|
| 3333 |
+
"loss": 0.3443,
|
| 3334 |
+
"step": 475
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 2.730401529636711,
|
| 3338 |
+
"grad_norm": 0.3562939165406292,
|
| 3339 |
+
"learning_rate": 4.709782476685476e-07,
|
| 3340 |
+
"loss": 0.393,
|
| 3341 |
+
"step": 476
|
| 3342 |
+
},
|
| 3343 |
+
{
|
| 3344 |
+
"epoch": 2.7361376673040154,
|
| 3345 |
+
"grad_norm": 0.3394312815950415,
|
| 3346 |
+
"learning_rate": 4.508771405844636e-07,
|
| 3347 |
+
"loss": 0.363,
|
| 3348 |
+
"step": 477
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 2.7418738049713194,
|
| 3352 |
+
"grad_norm": 0.3132735001523765,
|
| 3353 |
+
"learning_rate": 4.3120449905836746e-07,
|
| 3354 |
+
"loss": 0.2888,
|
| 3355 |
+
"step": 478
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 2.7476099426386233,
|
| 3359 |
+
"grad_norm": 0.35656415221445675,
|
| 3360 |
+
"learning_rate": 4.1196120579429786e-07,
|
| 3361 |
+
"loss": 0.369,
|
| 3362 |
+
"step": 479
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 2.7533460803059273,
|
| 3366 |
+
"grad_norm": 0.32703889632124866,
|
| 3367 |
+
"learning_rate": 3.931481242315993e-07,
|
| 3368 |
+
"loss": 0.3687,
|
| 3369 |
+
"step": 480
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 2.7590822179732313,
|
| 3373 |
+
"grad_norm": 0.3301574712781911,
|
| 3374 |
+
"learning_rate": 3.747660985061785e-07,
|
| 3375 |
+
"loss": 0.3511,
|
| 3376 |
+
"step": 481
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 2.7648183556405352,
|
| 3380 |
+
"grad_norm": 0.3445182321765397,
|
| 3381 |
+
"learning_rate": 3.5681595341263144e-07,
|
| 3382 |
+
"loss": 0.3887,
|
| 3383 |
+
"step": 482
|
| 3384 |
+
},
|
| 3385 |
+
{
|
| 3386 |
+
"epoch": 2.7705544933078396,
|
| 3387 |
+
"grad_norm": 0.35436336092040566,
|
| 3388 |
+
"learning_rate": 3.392984943672273e-07,
|
| 3389 |
+
"loss": 0.3865,
|
| 3390 |
+
"step": 483
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 2.7762906309751436,
|
| 3394 |
+
"grad_norm": 0.3384259106811793,
|
| 3395 |
+
"learning_rate": 3.2221450737178083e-07,
|
| 3396 |
+
"loss": 0.3501,
|
| 3397 |
+
"step": 484
|
| 3398 |
+
},
|
| 3399 |
+
{
|
| 3400 |
+
"epoch": 2.7820267686424476,
|
| 3401 |
+
"grad_norm": 0.3247213284472023,
|
| 3402 |
+
"learning_rate": 3.055647589783717e-07,
|
| 3403 |
+
"loss": 0.3548,
|
| 3404 |
+
"step": 485
|
| 3405 |
+
},
|
| 3406 |
+
{
|
| 3407 |
+
"epoch": 2.7877629063097515,
|
| 3408 |
+
"grad_norm": 0.3557473549812145,
|
| 3409 |
+
"learning_rate": 2.8934999625496287e-07,
|
| 3410 |
+
"loss": 0.3662,
|
| 3411 |
+
"step": 486
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 2.7934990439770555,
|
| 3415 |
+
"grad_norm": 0.3086740514405346,
|
| 3416 |
+
"learning_rate": 2.735709467518699e-07,
|
| 3417 |
+
"loss": 0.3535,
|
| 3418 |
+
"step": 487
|
| 3419 |
+
},
|
| 3420 |
+
{
|
| 3421 |
+
"epoch": 2.7992351816443595,
|
| 3422 |
+
"grad_norm": 0.332762209409699,
|
| 3423 |
+
"learning_rate": 2.5822831846912035e-07,
|
| 3424 |
+
"loss": 0.3995,
|
| 3425 |
+
"step": 488
|
| 3426 |
+
},
|
| 3427 |
+
{
|
| 3428 |
+
"epoch": 2.8049713193116634,
|
| 3429 |
+
"grad_norm": 0.3355698798094922,
|
| 3430 |
+
"learning_rate": 2.4332279982468454e-07,
|
| 3431 |
+
"loss": 0.35,
|
| 3432 |
+
"step": 489
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 2.8107074569789674,
|
| 3436 |
+
"grad_norm": 0.33988196442756863,
|
| 3437 |
+
"learning_rate": 2.2885505962359055e-07,
|
| 3438 |
+
"loss": 0.3727,
|
| 3439 |
+
"step": 490
|
| 3440 |
+
},
|
| 3441 |
+
{
|
| 3442 |
+
"epoch": 2.8164435946462714,
|
| 3443 |
+
"grad_norm": 0.3528168842787923,
|
| 3444 |
+
"learning_rate": 2.1482574702790804e-07,
|
| 3445 |
+
"loss": 0.3947,
|
| 3446 |
+
"step": 491
|
| 3447 |
+
},
|
| 3448 |
+
{
|
| 3449 |
+
"epoch": 2.8221797323135753,
|
| 3450 |
+
"grad_norm": 0.3440922584757431,
|
| 3451 |
+
"learning_rate": 2.0123549152762823e-07,
|
| 3452 |
+
"loss": 0.386,
|
| 3453 |
+
"step": 492
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 2.8279158699808793,
|
| 3457 |
+
"grad_norm": 0.34807725449747917,
|
| 3458 |
+
"learning_rate": 1.8808490291241433e-07,
|
| 3459 |
+
"loss": 0.4096,
|
| 3460 |
+
"step": 493
|
| 3461 |
+
},
|
| 3462 |
+
{
|
| 3463 |
+
"epoch": 2.8336520076481837,
|
| 3464 |
+
"grad_norm": 0.3305293457677881,
|
| 3465 |
+
"learning_rate": 1.7537457124423896e-07,
|
| 3466 |
+
"loss": 0.3414,
|
| 3467 |
+
"step": 494
|
| 3468 |
+
},
|
| 3469 |
+
{
|
| 3470 |
+
"epoch": 2.8393881453154877,
|
| 3471 |
+
"grad_norm": 0.36760723971309167,
|
| 3472 |
+
"learning_rate": 1.631050668309131e-07,
|
| 3473 |
+
"loss": 0.3896,
|
| 3474 |
+
"step": 495
|
| 3475 |
+
},
|
| 3476 |
+
{
|
| 3477 |
+
"epoch": 2.8451242829827916,
|
| 3478 |
+
"grad_norm": 0.31341641748111104,
|
| 3479 |
+
"learning_rate": 1.5127694020049432e-07,
|
| 3480 |
+
"loss": 0.3823,
|
| 3481 |
+
"step": 496
|
| 3482 |
+
},
|
| 3483 |
+
{
|
| 3484 |
+
"epoch": 2.8508604206500956,
|
| 3485 |
+
"grad_norm": 0.34699149322701534,
|
| 3486 |
+
"learning_rate": 1.3989072207658328e-07,
|
| 3487 |
+
"loss": 0.3725,
|
| 3488 |
+
"step": 497
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 2.8565965583173996,
|
| 3492 |
+
"grad_norm": 0.315099034771401,
|
| 3493 |
+
"learning_rate": 1.2894692335451376e-07,
|
| 3494 |
+
"loss": 0.2904,
|
| 3495 |
+
"step": 498
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 2.8623326959847035,
|
| 3499 |
+
"grad_norm": 0.3665483830887578,
|
| 3500 |
+
"learning_rate": 1.1844603507842667e-07,
|
| 3501 |
+
"loss": 0.4438,
|
| 3502 |
+
"step": 499
|
| 3503 |
+
},
|
| 3504 |
+
{
|
| 3505 |
+
"epoch": 2.8680688336520075,
|
| 3506 |
+
"grad_norm": 0.32419747257007453,
|
| 3507 |
+
"learning_rate": 1.0838852841923541e-07,
|
| 3508 |
+
"loss": 0.3444,
|
| 3509 |
+
"step": 500
|
| 3510 |
+
},
|
| 3511 |
+
{
|
| 3512 |
+
"epoch": 2.873804971319312,
|
| 3513 |
+
"grad_norm": 0.33442199544484646,
|
| 3514 |
+
"learning_rate": 9.877485465349057e-08,
|
| 3515 |
+
"loss": 0.3635,
|
| 3516 |
+
"step": 501
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 2.879541108986616,
|
| 3520 |
+
"grad_norm": 0.34177643056980267,
|
| 3521 |
+
"learning_rate": 8.960544514312275e-08,
|
| 3522 |
+
"loss": 0.3544,
|
| 3523 |
+
"step": 502
|
| 3524 |
+
},
|
| 3525 |
+
{
|
| 3526 |
+
"epoch": 2.88527724665392,
|
| 3527 |
+
"grad_norm": 0.3616719577615873,
|
| 3528 |
+
"learning_rate": 8.088071131609587e-08,
|
| 3529 |
+
"loss": 0.3823,
|
| 3530 |
+
"step": 503
|
| 3531 |
+
},
|
| 3532 |
+
{
|
| 3533 |
+
"epoch": 2.891013384321224,
|
| 3534 |
+
"grad_norm": 0.3351305518870644,
|
| 3535 |
+
"learning_rate": 7.260104464793971e-08,
|
| 3536 |
+
"loss": 0.3123,
|
| 3537 |
+
"step": 504
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 2.8967495219885278,
|
| 3541 |
+
"grad_norm": 0.31649142659834123,
|
| 3542 |
+
"learning_rate": 6.476681664419171e-08,
|
| 3543 |
+
"loss": 0.329,
|
| 3544 |
+
"step": 505
|
| 3545 |
+
},
|
| 3546 |
+
{
|
| 3547 |
+
"epoch": 2.9024856596558317,
|
| 3548 |
+
"grad_norm": 0.34029420103528796,
|
| 3549 |
+
"learning_rate": 5.737837882371921e-08,
|
| 3550 |
+
"loss": 0.4483,
|
| 3551 |
+
"step": 506
|
| 3552 |
+
},
|
| 3553 |
+
{
|
| 3554 |
+
"epoch": 2.9082217973231357,
|
| 3555 |
+
"grad_norm": 0.35364341776084457,
|
| 3556 |
+
"learning_rate": 5.0436062702956536e-08,
|
| 3557 |
+
"loss": 0.4067,
|
| 3558 |
+
"step": 507
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 2.9139579349904396,
|
| 3562 |
+
"grad_norm": 0.32827906264765444,
|
| 3563 |
+
"learning_rate": 4.394017978101905e-08,
|
| 3564 |
+
"loss": 0.3275,
|
| 3565 |
+
"step": 508
|
| 3566 |
+
},
|
| 3567 |
+
{
|
| 3568 |
+
"epoch": 2.9196940726577436,
|
| 3569 |
+
"grad_norm": 0.35463880013349575,
|
| 3570 |
+
"learning_rate": 3.789102152573665e-08,
|
| 3571 |
+
"loss": 0.3467,
|
| 3572 |
+
"step": 509
|
| 3573 |
+
},
|
| 3574 |
+
{
|
| 3575 |
+
"epoch": 2.9254302103250476,
|
| 3576 |
+
"grad_norm": 0.3610584326561177,
|
| 3577 |
+
"learning_rate": 3.228885936056858e-08,
|
| 3578 |
+
"loss": 0.3955,
|
| 3579 |
+
"step": 510
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 2.9311663479923515,
|
| 3583 |
+
"grad_norm": 0.337230906489088,
|
| 3584 |
+
"learning_rate": 2.7133944652429912e-08,
|
| 3585 |
+
"loss": 0.3531,
|
| 3586 |
+
"step": 511
|
| 3587 |
+
},
|
| 3588 |
+
{
|
| 3589 |
+
"epoch": 2.936902485659656,
|
| 3590 |
+
"grad_norm": 0.33412720856216777,
|
| 3591 |
+
"learning_rate": 2.242650870040497e-08,
|
| 3592 |
+
"loss": 0.368,
|
| 3593 |
+
"step": 512
|
| 3594 |
+
},
|
| 3595 |
+
{
|
| 3596 |
+
"epoch": 2.94263862332696,
|
| 3597 |
+
"grad_norm": 0.3286692837778851,
|
| 3598 |
+
"learning_rate": 1.8166762725381205e-08,
|
| 3599 |
+
"loss": 0.3218,
|
| 3600 |
+
"step": 513
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 2.948374760994264,
|
| 3604 |
+
"grad_norm": 0.32866930833280994,
|
| 3605 |
+
"learning_rate": 1.4354897860558992e-08,
|
| 3606 |
+
"loss": 0.3885,
|
| 3607 |
+
"step": 514
|
| 3608 |
+
},
|
| 3609 |
+
{
|
| 3610 |
+
"epoch": 2.954110898661568,
|
| 3611 |
+
"grad_norm": 0.34657396795048584,
|
| 3612 |
+
"learning_rate": 1.099108514288627e-08,
|
| 3613 |
+
"loss": 0.3649,
|
| 3614 |
+
"step": 515
|
| 3615 |
+
},
|
| 3616 |
+
{
|
| 3617 |
+
"epoch": 2.959847036328872,
|
| 3618 |
+
"grad_norm": 0.36485883198814917,
|
| 3619 |
+
"learning_rate": 8.075475505373576e-09,
|
| 3620 |
+
"loss": 0.3413,
|
| 3621 |
+
"step": 516
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 2.9655831739961758,
|
| 3625 |
+
"grad_norm": 0.34064688921305847,
|
| 3626 |
+
"learning_rate": 5.608199770334999e-09,
|
| 3627 |
+
"loss": 0.337,
|
| 3628 |
+
"step": 517
|
| 3629 |
+
},
|
| 3630 |
+
{
|
| 3631 |
+
"epoch": 2.97131931166348,
|
| 3632 |
+
"grad_norm": 0.34881696844494053,
|
| 3633 |
+
"learning_rate": 3.5893686435028995e-09,
|
| 3634 |
+
"loss": 0.4078,
|
| 3635 |
+
"step": 518
|
| 3636 |
+
},
|
| 3637 |
+
{
|
| 3638 |
+
"epoch": 2.977055449330784,
|
| 3639 |
+
"grad_norm": 0.3403683643532366,
|
| 3640 |
+
"learning_rate": 2.019072709074088e-09,
|
| 3641 |
+
"loss": 0.3268,
|
| 3642 |
+
"step": 519
|
| 3643 |
+
},
|
| 3644 |
+
{
|
| 3645 |
+
"epoch": 2.982791586998088,
|
| 3646 |
+
"grad_norm": 0.3337616239063328,
|
| 3647 |
+
"learning_rate": 8.973824256364172e-10,
|
| 3648 |
+
"loss": 0.3836,
|
| 3649 |
+
"step": 520
|
| 3650 |
+
},
|
| 3651 |
+
{
|
| 3652 |
+
"epoch": 2.988527724665392,
|
| 3653 |
+
"grad_norm": 0.34240634016525656,
|
| 3654 |
+
"learning_rate": 2.2434812301352915e-10,
|
| 3655 |
+
"loss": 0.3802,
|
| 3656 |
+
"step": 521
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 2.994263862332696,
|
| 3660 |
+
"grad_norm": 0.3228196417505892,
|
| 3661 |
+
"learning_rate": 0.0,
|
| 3662 |
+
"loss": 0.3441,
|
| 3663 |
+
"step": 522
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 2.994263862332696,
|
| 3667 |
+
"step": 522,
|
| 3668 |
+
"total_flos": 5.620362022127534e+17,
|
| 3669 |
+
"train_loss": 0.45750690551324824,
|
| 3670 |
+
"train_runtime": 8551.5314,
|
| 3671 |
+
"train_samples_per_second": 5.862,
|
| 3672 |
+
"train_steps_per_second": 0.061
|
| 3673 |
+
}
|
| 3674 |
+
],
|
| 3675 |
+
"logging_steps": 1.0,
|
| 3676 |
+
"max_steps": 522,
|
| 3677 |
+
"num_input_tokens_seen": 0,
|
| 3678 |
+
"num_train_epochs": 3,
|
| 3679 |
+
"save_steps": 500,
|
| 3680 |
+
"stateful_callbacks": {
|
| 3681 |
+
"TrainerControl": {
|
| 3682 |
+
"args": {
|
| 3683 |
+
"should_epoch_stop": false,
|
| 3684 |
+
"should_evaluate": false,
|
| 3685 |
+
"should_log": false,
|
| 3686 |
+
"should_save": true,
|
| 3687 |
+
"should_training_stop": true
|
| 3688 |
+
},
|
| 3689 |
+
"attributes": {}
|
| 3690 |
+
}
|
| 3691 |
+
},
|
| 3692 |
+
"total_flos": 5.620362022127534e+17,
|
| 3693 |
+
"train_batch_size": 1,
|
| 3694 |
+
"trial_name": null,
|
| 3695 |
+
"trial_params": null
|
| 3696 |
+
}
|
training_loss.png
ADDED
|