End of training
Browse files- README.md +3 -2
- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +2292 -0
- training_loss.png +0 -0
- training_rewards_accuracies.png +0 -0
README.md
CHANGED
|
@@ -3,9 +3,10 @@ library_name: peft
|
|
| 3 |
license: llama3
|
| 4 |
base_model: meta-llama/Meta-Llama-3-8B-Instruct
|
| 5 |
tags:
|
|
|
|
|
|
|
| 6 |
- trl
|
| 7 |
- dpo
|
| 8 |
-
- llama-factory
|
| 9 |
- generated_from_trainer
|
| 10 |
model-index:
|
| 11 |
- name: dpo_crafting_lora
|
|
@@ -17,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 17 |
|
| 18 |
# dpo_crafting_lora
|
| 19 |
|
| 20 |
-
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) on
|
| 21 |
|
| 22 |
## Model description
|
| 23 |
|
|
|
|
| 3 |
license: llama3
|
| 4 |
base_model: meta-llama/Meta-Llama-3-8B-Instruct
|
| 5 |
tags:
|
| 6 |
+
- llama-factory
|
| 7 |
+
- lora
|
| 8 |
- trl
|
| 9 |
- dpo
|
|
|
|
| 10 |
- generated_from_trainer
|
| 11 |
model-index:
|
| 12 |
- name: dpo_crafting_lora
|
|
|
|
| 18 |
|
| 19 |
# dpo_crafting_lora
|
| 20 |
|
| 21 |
+
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) on the crafting_dpo_data dataset.
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.93687707641196,
|
| 3 |
+
"total_flos": 2.9219422466201354e+19,
|
| 4 |
+
"train_loss": 0.07423472365736962,
|
| 5 |
+
"train_runtime": 24807.8996,
|
| 6 |
+
"train_samples_per_second": 3.881,
|
| 7 |
+
"train_steps_per_second": 0.06
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.93687707641196,
|
| 3 |
+
"total_flos": 2.9219422466201354e+19,
|
| 4 |
+
"train_loss": 0.07423472365736962,
|
| 5 |
+
"train_runtime": 24807.8996,
|
| 6 |
+
"train_samples_per_second": 3.881,
|
| 7 |
+
"train_steps_per_second": 0.06
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2292 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 9.93687707641196,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1500,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0664451827242525,
|
| 13 |
+
"grad_norm": 40.48354721069336,
|
| 14 |
+
"learning_rate": 3.3333333333333335e-07,
|
| 15 |
+
"logits/chosen": -0.12491344660520554,
|
| 16 |
+
"logits/rejected": -0.18818695843219757,
|
| 17 |
+
"logps/chosen": -1117.441650390625,
|
| 18 |
+
"logps/rejected": -1311.48095703125,
|
| 19 |
+
"loss": 0.726,
|
| 20 |
+
"rewards/accuracies": 0.3765625059604645,
|
| 21 |
+
"rewards/chosen": -0.022913634777069092,
|
| 22 |
+
"rewards/margins": -0.004669740330427885,
|
| 23 |
+
"rewards/rejected": -0.018243903294205666,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.132890365448505,
|
| 28 |
+
"grad_norm": 68.69932556152344,
|
| 29 |
+
"learning_rate": 6.666666666666667e-07,
|
| 30 |
+
"logits/chosen": -0.12522226572036743,
|
| 31 |
+
"logits/rejected": -0.18426118791103363,
|
| 32 |
+
"logps/chosen": -1105.445068359375,
|
| 33 |
+
"logps/rejected": -1318.8582763671875,
|
| 34 |
+
"loss": 0.7146,
|
| 35 |
+
"rewards/accuracies": 0.504687488079071,
|
| 36 |
+
"rewards/chosen": -0.014212612062692642,
|
| 37 |
+
"rewards/margins": 0.031459741294384,
|
| 38 |
+
"rewards/rejected": -0.04567235708236694,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.19933554817275748,
|
| 43 |
+
"grad_norm": 32.92807388305664,
|
| 44 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 45 |
+
"logits/chosen": -0.13368168473243713,
|
| 46 |
+
"logits/rejected": -0.19534170627593994,
|
| 47 |
+
"logps/chosen": -1098.4427490234375,
|
| 48 |
+
"logps/rejected": -1286.0972900390625,
|
| 49 |
+
"loss": 0.7014,
|
| 50 |
+
"rewards/accuracies": 0.5453125238418579,
|
| 51 |
+
"rewards/chosen": -0.14931853115558624,
|
| 52 |
+
"rewards/margins": 0.06388694047927856,
|
| 53 |
+
"rewards/rejected": -0.2132054567337036,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.26578073089701,
|
| 58 |
+
"grad_norm": 35.1468391418457,
|
| 59 |
+
"learning_rate": 1.3333333333333334e-06,
|
| 60 |
+
"logits/chosen": -0.13665804266929626,
|
| 61 |
+
"logits/rejected": -0.19470006227493286,
|
| 62 |
+
"logps/chosen": -1057.49560546875,
|
| 63 |
+
"logps/rejected": -1253.0081787109375,
|
| 64 |
+
"loss": 0.6609,
|
| 65 |
+
"rewards/accuracies": 0.5296874642372131,
|
| 66 |
+
"rewards/chosen": -0.42790666222572327,
|
| 67 |
+
"rewards/margins": 0.21245712041854858,
|
| 68 |
+
"rewards/rejected": -0.6403638124465942,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.33222591362126247,
|
| 73 |
+
"grad_norm": 23.31159210205078,
|
| 74 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 75 |
+
"logits/chosen": -0.1325100213289261,
|
| 76 |
+
"logits/rejected": -0.19247327744960785,
|
| 77 |
+
"logps/chosen": -1068.646484375,
|
| 78 |
+
"logps/rejected": -1261.2486572265625,
|
| 79 |
+
"loss": 0.6518,
|
| 80 |
+
"rewards/accuracies": 0.596875011920929,
|
| 81 |
+
"rewards/chosen": -0.8173043131828308,
|
| 82 |
+
"rewards/margins": 0.3520071506500244,
|
| 83 |
+
"rewards/rejected": -1.1693115234375,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.39867109634551495,
|
| 88 |
+
"grad_norm": 21.35710334777832,
|
| 89 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 90 |
+
"logits/chosen": -0.1356111764907837,
|
| 91 |
+
"logits/rejected": -0.19784294068813324,
|
| 92 |
+
"logps/chosen": -1058.8770751953125,
|
| 93 |
+
"logps/rejected": -1271.4691162109375,
|
| 94 |
+
"loss": 0.6231,
|
| 95 |
+
"rewards/accuracies": 0.6468750238418579,
|
| 96 |
+
"rewards/chosen": -0.9412558078765869,
|
| 97 |
+
"rewards/margins": 0.5320442914962769,
|
| 98 |
+
"rewards/rejected": -1.4733000993728638,
|
| 99 |
+
"step": 60
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.46511627906976744,
|
| 103 |
+
"grad_norm": 23.317035675048828,
|
| 104 |
+
"learning_rate": 2.3333333333333336e-06,
|
| 105 |
+
"logits/chosen": -0.13605119287967682,
|
| 106 |
+
"logits/rejected": -0.19660170376300812,
|
| 107 |
+
"logps/chosen": -1103.0458984375,
|
| 108 |
+
"logps/rejected": -1314.48291015625,
|
| 109 |
+
"loss": 0.5443,
|
| 110 |
+
"rewards/accuracies": 0.7125000357627869,
|
| 111 |
+
"rewards/chosen": -0.8000861406326294,
|
| 112 |
+
"rewards/margins": 0.6938734650611877,
|
| 113 |
+
"rewards/rejected": -1.493959665298462,
|
| 114 |
+
"step": 70
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.53156146179402,
|
| 118 |
+
"grad_norm": 22.587018966674805,
|
| 119 |
+
"learning_rate": 2.666666666666667e-06,
|
| 120 |
+
"logits/chosen": -0.11189720779657364,
|
| 121 |
+
"logits/rejected": -0.18983830511569977,
|
| 122 |
+
"logps/chosen": -1066.6973876953125,
|
| 123 |
+
"logps/rejected": -1319.9337158203125,
|
| 124 |
+
"loss": 0.4977,
|
| 125 |
+
"rewards/accuracies": 0.7718749642372131,
|
| 126 |
+
"rewards/chosen": -0.792549729347229,
|
| 127 |
+
"rewards/margins": 1.0354700088500977,
|
| 128 |
+
"rewards/rejected": -1.8280198574066162,
|
| 129 |
+
"step": 80
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.5980066445182725,
|
| 133 |
+
"grad_norm": 20.043182373046875,
|
| 134 |
+
"learning_rate": 3e-06,
|
| 135 |
+
"logits/chosen": -0.1244162917137146,
|
| 136 |
+
"logits/rejected": -0.188711017370224,
|
| 137 |
+
"logps/chosen": -1076.73779296875,
|
| 138 |
+
"logps/rejected": -1331.7142333984375,
|
| 139 |
+
"loss": 0.4294,
|
| 140 |
+
"rewards/accuracies": 0.7984374761581421,
|
| 141 |
+
"rewards/chosen": -0.42743515968322754,
|
| 142 |
+
"rewards/margins": 1.2227166891098022,
|
| 143 |
+
"rewards/rejected": -1.6501517295837402,
|
| 144 |
+
"step": 90
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.6644518272425249,
|
| 148 |
+
"grad_norm": 10.029542922973633,
|
| 149 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 150 |
+
"logits/chosen": -0.12235234677791595,
|
| 151 |
+
"logits/rejected": -0.1966520994901657,
|
| 152 |
+
"logps/chosen": -1058.425537109375,
|
| 153 |
+
"logps/rejected": -1334.389404296875,
|
| 154 |
+
"loss": 0.3671,
|
| 155 |
+
"rewards/accuracies": 0.871874988079071,
|
| 156 |
+
"rewards/chosen": -0.5026114583015442,
|
| 157 |
+
"rewards/margins": 1.8994375467300415,
|
| 158 |
+
"rewards/rejected": -2.4020493030548096,
|
| 159 |
+
"step": 100
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.7308970099667774,
|
| 163 |
+
"grad_norm": 11.281858444213867,
|
| 164 |
+
"learning_rate": 3.6666666666666666e-06,
|
| 165 |
+
"logits/chosen": -0.13173305988311768,
|
| 166 |
+
"logits/rejected": -0.19946981966495514,
|
| 167 |
+
"logps/chosen": -1074.1378173828125,
|
| 168 |
+
"logps/rejected": -1311.911865234375,
|
| 169 |
+
"loss": 0.2999,
|
| 170 |
+
"rewards/accuracies": 0.895312488079071,
|
| 171 |
+
"rewards/chosen": -0.37171292304992676,
|
| 172 |
+
"rewards/margins": 2.6020548343658447,
|
| 173 |
+
"rewards/rejected": -2.9737677574157715,
|
| 174 |
+
"step": 110
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.7973421926910299,
|
| 178 |
+
"grad_norm": 11.228713989257812,
|
| 179 |
+
"learning_rate": 4.000000000000001e-06,
|
| 180 |
+
"logits/chosen": -0.12718108296394348,
|
| 181 |
+
"logits/rejected": -0.19788110256195068,
|
| 182 |
+
"logps/chosen": -1091.0115966796875,
|
| 183 |
+
"logps/rejected": -1348.0794677734375,
|
| 184 |
+
"loss": 0.2278,
|
| 185 |
+
"rewards/accuracies": 0.9249999523162842,
|
| 186 |
+
"rewards/chosen": -0.0037933550775051117,
|
| 187 |
+
"rewards/margins": 4.05342435836792,
|
| 188 |
+
"rewards/rejected": -4.057217597961426,
|
| 189 |
+
"step": 120
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.8637873754152824,
|
| 193 |
+
"grad_norm": 7.068265438079834,
|
| 194 |
+
"learning_rate": 4.333333333333334e-06,
|
| 195 |
+
"logits/chosen": -0.12988580763339996,
|
| 196 |
+
"logits/rejected": -0.19472359120845795,
|
| 197 |
+
"logps/chosen": -1101.286865234375,
|
| 198 |
+
"logps/rejected": -1364.6812744140625,
|
| 199 |
+
"loss": 0.162,
|
| 200 |
+
"rewards/accuracies": 0.9515625834465027,
|
| 201 |
+
"rewards/chosen": 0.24117425084114075,
|
| 202 |
+
"rewards/margins": 5.719590663909912,
|
| 203 |
+
"rewards/rejected": -5.4784159660339355,
|
| 204 |
+
"step": 130
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.9302325581395349,
|
| 208 |
+
"grad_norm": 2.4044954776763916,
|
| 209 |
+
"learning_rate": 4.666666666666667e-06,
|
| 210 |
+
"logits/chosen": -0.13687601685523987,
|
| 211 |
+
"logits/rejected": -0.20584635436534882,
|
| 212 |
+
"logps/chosen": -1033.876708984375,
|
| 213 |
+
"logps/rejected": -1361.149658203125,
|
| 214 |
+
"loss": 0.1267,
|
| 215 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 216 |
+
"rewards/chosen": 1.0959956645965576,
|
| 217 |
+
"rewards/margins": 7.263400077819824,
|
| 218 |
+
"rewards/rejected": -6.1674041748046875,
|
| 219 |
+
"step": 140
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.9966777408637874,
|
| 223 |
+
"grad_norm": 23.234800338745117,
|
| 224 |
+
"learning_rate": 5e-06,
|
| 225 |
+
"logits/chosen": -0.14618705213069916,
|
| 226 |
+
"logits/rejected": -0.2165984809398651,
|
| 227 |
+
"logps/chosen": -1056.0648193359375,
|
| 228 |
+
"logps/rejected": -1392.6947021484375,
|
| 229 |
+
"loss": 0.1071,
|
| 230 |
+
"rewards/accuracies": 0.9390624761581421,
|
| 231 |
+
"rewards/chosen": 0.6555425524711609,
|
| 232 |
+
"rewards/margins": 9.398031234741211,
|
| 233 |
+
"rewards/rejected": -8.742487907409668,
|
| 234 |
+
"step": 150
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 1.0598006644518272,
|
| 238 |
+
"grad_norm": 5.546257495880127,
|
| 239 |
+
"learning_rate": 4.999323102948655e-06,
|
| 240 |
+
"logits/chosen": -0.1447768211364746,
|
| 241 |
+
"logits/rejected": -0.2160787135362625,
|
| 242 |
+
"logps/chosen": -1052.21728515625,
|
| 243 |
+
"logps/rejected": -1390.3724365234375,
|
| 244 |
+
"loss": 0.084,
|
| 245 |
+
"rewards/accuracies": 0.9391447901725769,
|
| 246 |
+
"rewards/chosen": 1.6069414615631104,
|
| 247 |
+
"rewards/margins": 11.616178512573242,
|
| 248 |
+
"rewards/rejected": -10.009236335754395,
|
| 249 |
+
"step": 160
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.1262458471760797,
|
| 253 |
+
"grad_norm": 29.2595157623291,
|
| 254 |
+
"learning_rate": 4.997292778346312e-06,
|
| 255 |
+
"logits/chosen": -0.15423689782619476,
|
| 256 |
+
"logits/rejected": -0.21319061517715454,
|
| 257 |
+
"logps/chosen": -1064.402587890625,
|
| 258 |
+
"logps/rejected": -1432.095458984375,
|
| 259 |
+
"loss": 0.061,
|
| 260 |
+
"rewards/accuracies": 0.9578124284744263,
|
| 261 |
+
"rewards/chosen": 1.6962934732437134,
|
| 262 |
+
"rewards/margins": 15.001029968261719,
|
| 263 |
+
"rewards/rejected": -13.304736137390137,
|
| 264 |
+
"step": 170
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 1.1926910299003322,
|
| 268 |
+
"grad_norm": 3.240455150604248,
|
| 269 |
+
"learning_rate": 4.993910125649561e-06,
|
| 270 |
+
"logits/chosen": -0.13828535377979279,
|
| 271 |
+
"logits/rejected": -0.21751590073108673,
|
| 272 |
+
"logps/chosen": -998.13134765625,
|
| 273 |
+
"logps/rejected": -1392.8687744140625,
|
| 274 |
+
"loss": 0.0494,
|
| 275 |
+
"rewards/accuracies": 0.9625000357627869,
|
| 276 |
+
"rewards/chosen": 2.9092648029327393,
|
| 277 |
+
"rewards/margins": 15.800436019897461,
|
| 278 |
+
"rewards/rejected": -12.891172409057617,
|
| 279 |
+
"step": 180
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 1.2591362126245846,
|
| 283 |
+
"grad_norm": 0.4213032126426697,
|
| 284 |
+
"learning_rate": 4.989176976624511e-06,
|
| 285 |
+
"logits/chosen": -0.16740064322948456,
|
| 286 |
+
"logits/rejected": -0.23276962339878082,
|
| 287 |
+
"logps/chosen": -1051.3272705078125,
|
| 288 |
+
"logps/rejected": -1436.19189453125,
|
| 289 |
+
"loss": 0.0566,
|
| 290 |
+
"rewards/accuracies": 0.942187488079071,
|
| 291 |
+
"rewards/chosen": 3.340867757797241,
|
| 292 |
+
"rewards/margins": 18.916629791259766,
|
| 293 |
+
"rewards/rejected": -15.575761795043945,
|
| 294 |
+
"step": 190
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 1.3255813953488373,
|
| 298 |
+
"grad_norm": 33.59416961669922,
|
| 299 |
+
"learning_rate": 4.983095894354858e-06,
|
| 300 |
+
"logits/chosen": -0.17280790209770203,
|
| 301 |
+
"logits/rejected": -0.23289522528648376,
|
| 302 |
+
"logps/chosen": -1036.917236328125,
|
| 303 |
+
"logps/rejected": -1430.266357421875,
|
| 304 |
+
"loss": 0.0552,
|
| 305 |
+
"rewards/accuracies": 0.9453125,
|
| 306 |
+
"rewards/chosen": 4.887056827545166,
|
| 307 |
+
"rewards/margins": 18.91439437866211,
|
| 308 |
+
"rewards/rejected": -14.027338027954102,
|
| 309 |
+
"step": 200
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 1.3920265780730898,
|
| 313 |
+
"grad_norm": 0.12995536625385284,
|
| 314 |
+
"learning_rate": 4.975670171853926e-06,
|
| 315 |
+
"logits/chosen": -0.16127409040927887,
|
| 316 |
+
"logits/rejected": -0.22987647354602814,
|
| 317 |
+
"logps/chosen": -1064.2008056640625,
|
| 318 |
+
"logps/rejected": -1541.5230712890625,
|
| 319 |
+
"loss": 0.0741,
|
| 320 |
+
"rewards/accuracies": 0.9515624642372131,
|
| 321 |
+
"rewards/chosen": 1.9763206243515015,
|
| 322 |
+
"rewards/margins": 24.5360107421875,
|
| 323 |
+
"rewards/rejected": -22.5596923828125,
|
| 324 |
+
"step": 210
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.4584717607973423,
|
| 328 |
+
"grad_norm": 125.04955291748047,
|
| 329 |
+
"learning_rate": 4.966903830281449e-06,
|
| 330 |
+
"logits/chosen": -0.16986006498336792,
|
| 331 |
+
"logits/rejected": -0.23232412338256836,
|
| 332 |
+
"logps/chosen": -1011.2138061523438,
|
| 333 |
+
"logps/rejected": -1458.410888671875,
|
| 334 |
+
"loss": 0.0702,
|
| 335 |
+
"rewards/accuracies": 0.9468750357627869,
|
| 336 |
+
"rewards/chosen": 4.205624580383301,
|
| 337 |
+
"rewards/margins": 22.20413589477539,
|
| 338 |
+
"rewards/rejected": -17.998512268066406,
|
| 339 |
+
"step": 220
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.5249169435215948,
|
| 343 |
+
"grad_norm": 1.0301318168640137,
|
| 344 |
+
"learning_rate": 4.956801616766033e-06,
|
| 345 |
+
"logits/chosen": -0.17181265354156494,
|
| 346 |
+
"logits/rejected": -0.22897139191627502,
|
| 347 |
+
"logps/chosen": -1064.4287109375,
|
| 348 |
+
"logps/rejected": -1548.607177734375,
|
| 349 |
+
"loss": 0.0419,
|
| 350 |
+
"rewards/accuracies": 0.956250011920929,
|
| 351 |
+
"rewards/chosen": 4.201747894287109,
|
| 352 |
+
"rewards/margins": 26.632787704467773,
|
| 353 |
+
"rewards/rejected": -22.43103790283203,
|
| 354 |
+
"step": 230
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.5913621262458473,
|
| 358 |
+
"grad_norm": 1.9554957151412964,
|
| 359 |
+
"learning_rate": 4.9453690018345144e-06,
|
| 360 |
+
"logits/chosen": -0.17581965029239655,
|
| 361 |
+
"logits/rejected": -0.23298974335193634,
|
| 362 |
+
"logps/chosen": -1037.4566650390625,
|
| 363 |
+
"logps/rejected": -1497.9283447265625,
|
| 364 |
+
"loss": 0.0354,
|
| 365 |
+
"rewards/accuracies": 0.957812488079071,
|
| 366 |
+
"rewards/chosen": 4.2116312980651855,
|
| 367 |
+
"rewards/margins": 25.39874267578125,
|
| 368 |
+
"rewards/rejected": -21.18711280822754,
|
| 369 |
+
"step": 240
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.6578073089700998,
|
| 373 |
+
"grad_norm": 0.35193875432014465,
|
| 374 |
+
"learning_rate": 4.93261217644956e-06,
|
| 375 |
+
"logits/chosen": -0.16469064354896545,
|
| 376 |
+
"logits/rejected": -0.23437215387821198,
|
| 377 |
+
"logps/chosen": -1052.7650146484375,
|
| 378 |
+
"logps/rejected": -1553.04833984375,
|
| 379 |
+
"loss": 0.0423,
|
| 380 |
+
"rewards/accuracies": 0.96875,
|
| 381 |
+
"rewards/chosen": 4.150611877441406,
|
| 382 |
+
"rewards/margins": 26.79295539855957,
|
| 383 |
+
"rewards/rejected": -22.642345428466797,
|
| 384 |
+
"step": 250
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.7242524916943522,
|
| 388 |
+
"grad_norm": 0.6141932606697083,
|
| 389 |
+
"learning_rate": 4.91853804865716e-06,
|
| 390 |
+
"logits/chosen": -0.18150699138641357,
|
| 391 |
+
"logits/rejected": -0.2411101907491684,
|
| 392 |
+
"logps/chosen": -972.6925048828125,
|
| 393 |
+
"logps/rejected": -1363.6278076171875,
|
| 394 |
+
"loss": 0.0335,
|
| 395 |
+
"rewards/accuracies": 0.9609375,
|
| 396 |
+
"rewards/chosen": 6.711765289306641,
|
| 397 |
+
"rewards/margins": 22.288022994995117,
|
| 398 |
+
"rewards/rejected": -15.576257705688477,
|
| 399 |
+
"step": 260
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.7906976744186047,
|
| 403 |
+
"grad_norm": 36.097991943359375,
|
| 404 |
+
"learning_rate": 4.903154239845798e-06,
|
| 405 |
+
"logits/chosen": -0.17819088697433472,
|
| 406 |
+
"logits/rejected": -0.2480602264404297,
|
| 407 |
+
"logps/chosen": -1043.45751953125,
|
| 408 |
+
"logps/rejected": -1562.9696044921875,
|
| 409 |
+
"loss": 0.0432,
|
| 410 |
+
"rewards/accuracies": 0.9500000476837158,
|
| 411 |
+
"rewards/chosen": 4.969973087310791,
|
| 412 |
+
"rewards/margins": 29.227781295776367,
|
| 413 |
+
"rewards/rejected": -24.25780487060547,
|
| 414 |
+
"step": 270
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 1.8571428571428572,
|
| 418 |
+
"grad_norm": 0.804478645324707,
|
| 419 |
+
"learning_rate": 4.88646908061933e-06,
|
| 420 |
+
"logits/chosen": -0.18135662376880646,
|
| 421 |
+
"logits/rejected": -0.2500635087490082,
|
| 422 |
+
"logps/chosen": -973.140869140625,
|
| 423 |
+
"logps/rejected": -1519.1575927734375,
|
| 424 |
+
"loss": 0.0387,
|
| 425 |
+
"rewards/accuracies": 0.948437511920929,
|
| 426 |
+
"rewards/chosen": 6.344993591308594,
|
| 427 |
+
"rewards/margins": 28.498210906982422,
|
| 428 |
+
"rewards/rejected": -22.15321922302246,
|
| 429 |
+
"step": 280
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.9235880398671097,
|
| 433 |
+
"grad_norm": 0.21806667745113373,
|
| 434 |
+
"learning_rate": 4.868491606285823e-06,
|
| 435 |
+
"logits/chosen": -0.1847972869873047,
|
| 436 |
+
"logits/rejected": -0.24828991293907166,
|
| 437 |
+
"logps/chosen": -1015.7157592773438,
|
| 438 |
+
"logps/rejected": -1512.730712890625,
|
| 439 |
+
"loss": 0.0349,
|
| 440 |
+
"rewards/accuracies": 0.9546875357627869,
|
| 441 |
+
"rewards/chosen": 7.247753620147705,
|
| 442 |
+
"rewards/margins": 27.996492385864258,
|
| 443 |
+
"rewards/rejected": -20.748737335205078,
|
| 444 |
+
"step": 290
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.9900332225913622,
|
| 448 |
+
"grad_norm": 0.06243917718529701,
|
| 449 |
+
"learning_rate": 4.849231551964771e-06,
|
| 450 |
+
"logits/chosen": -0.18840095400810242,
|
| 451 |
+
"logits/rejected": -0.2502696216106415,
|
| 452 |
+
"logps/chosen": -1035.1356201171875,
|
| 453 |
+
"logps/rejected": -1553.83154296875,
|
| 454 |
+
"loss": 0.0338,
|
| 455 |
+
"rewards/accuracies": 0.9562500715255737,
|
| 456 |
+
"rewards/chosen": 6.608232021331787,
|
| 457 |
+
"rewards/margins": 30.3477783203125,
|
| 458 |
+
"rewards/rejected": -23.739547729492188,
|
| 459 |
+
"step": 300
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 2.053156146179402,
|
| 463 |
+
"grad_norm": 1.4525890350341797,
|
| 464 |
+
"learning_rate": 4.828699347315357e-06,
|
| 465 |
+
"logits/chosen": -0.16825956106185913,
|
| 466 |
+
"logits/rejected": -0.2412978559732437,
|
| 467 |
+
"logps/chosen": -993.029541015625,
|
| 468 |
+
"logps/rejected": -1512.5572509765625,
|
| 469 |
+
"loss": 0.0283,
|
| 470 |
+
"rewards/accuracies": 0.9621711373329163,
|
| 471 |
+
"rewards/chosen": 7.502988815307617,
|
| 472 |
+
"rewards/margins": 29.106731414794922,
|
| 473 |
+
"rewards/rejected": -21.603742599487305,
|
| 474 |
+
"step": 310
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 2.1196013289036544,
|
| 478 |
+
"grad_norm": 0.341370165348053,
|
| 479 |
+
"learning_rate": 4.806906110888606e-06,
|
| 480 |
+
"logits/chosen": -0.19253456592559814,
|
| 481 |
+
"logits/rejected": -0.24857579171657562,
|
| 482 |
+
"logps/chosen": -1003.42822265625,
|
| 483 |
+
"logps/rejected": -1490.8272705078125,
|
| 484 |
+
"loss": 0.0341,
|
| 485 |
+
"rewards/accuracies": 0.953125,
|
| 486 |
+
"rewards/chosen": 6.586777210235596,
|
| 487 |
+
"rewards/margins": 30.160865783691406,
|
| 488 |
+
"rewards/rejected": -23.574087142944336,
|
| 489 |
+
"step": 320
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 2.186046511627907,
|
| 493 |
+
"grad_norm": 0.32293370366096497,
|
| 494 |
+
"learning_rate": 4.783863644106502e-06,
|
| 495 |
+
"logits/chosen": -0.18394093215465546,
|
| 496 |
+
"logits/rejected": -0.2500993013381958,
|
| 497 |
+
"logps/chosen": -998.8344116210938,
|
| 498 |
+
"logps/rejected": -1539.57958984375,
|
| 499 |
+
"loss": 0.0293,
|
| 500 |
+
"rewards/accuracies": 0.9593749642372131,
|
| 501 |
+
"rewards/chosen": 6.540307521820068,
|
| 502 |
+
"rewards/margins": 32.438438415527344,
|
| 503 |
+
"rewards/rejected": -25.898130416870117,
|
| 504 |
+
"step": 330
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 2.2524916943521593,
|
| 508 |
+
"grad_norm": 0.4895381033420563,
|
| 509 |
+
"learning_rate": 4.759584424871302e-06,
|
| 510 |
+
"logits/chosen": -0.19156453013420105,
|
| 511 |
+
"logits/rejected": -0.2491527646780014,
|
| 512 |
+
"logps/chosen": -1053.5201416015625,
|
| 513 |
+
"logps/rejected": -1579.8900146484375,
|
| 514 |
+
"loss": 0.0266,
|
| 515 |
+
"rewards/accuracies": 0.964062511920929,
|
| 516 |
+
"rewards/chosen": 8.41015338897705,
|
| 517 |
+
"rewards/margins": 31.691499710083008,
|
| 518 |
+
"rewards/rejected": -23.281347274780273,
|
| 519 |
+
"step": 340
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 2.318936877076412,
|
| 523 |
+
"grad_norm": 0.6678661108016968,
|
| 524 |
+
"learning_rate": 4.734081600808531e-06,
|
| 525 |
+
"logits/chosen": -0.2005804479122162,
|
| 526 |
+
"logits/rejected": -0.2595979571342468,
|
| 527 |
+
"logps/chosen": -1060.26025390625,
|
| 528 |
+
"logps/rejected": -1603.9803466796875,
|
| 529 |
+
"loss": 0.0389,
|
| 530 |
+
"rewards/accuracies": 0.9468750357627869,
|
| 531 |
+
"rewards/chosen": 7.6117658615112305,
|
| 532 |
+
"rewards/margins": 33.64573287963867,
|
| 533 |
+
"rewards/rejected": -26.03396987915039,
|
| 534 |
+
"step": 350
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 2.3853820598006643,
|
| 538 |
+
"grad_norm": 0.05852515622973442,
|
| 539 |
+
"learning_rate": 4.707368982147318e-06,
|
| 540 |
+
"logits/chosen": -0.18053898215293884,
|
| 541 |
+
"logits/rejected": -0.25306934118270874,
|
| 542 |
+
"logps/chosen": -965.810791015625,
|
| 543 |
+
"logps/rejected": -1501.8775634765625,
|
| 544 |
+
"loss": 0.0405,
|
| 545 |
+
"rewards/accuracies": 0.9578125476837158,
|
| 546 |
+
"rewards/chosen": 7.657576560974121,
|
| 547 |
+
"rewards/margins": 30.364038467407227,
|
| 548 |
+
"rewards/rejected": -22.706459045410156,
|
| 549 |
+
"step": 360
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 2.451827242524917,
|
| 553 |
+
"grad_norm": 0.2221815437078476,
|
| 554 |
+
"learning_rate": 4.679461034241906e-06,
|
| 555 |
+
"logits/chosen": -0.1780516505241394,
|
| 556 |
+
"logits/rejected": -0.25180700421333313,
|
| 557 |
+
"logps/chosen": -1041.829345703125,
|
| 558 |
+
"logps/rejected": -1607.8099365234375,
|
| 559 |
+
"loss": 0.0274,
|
| 560 |
+
"rewards/accuracies": 0.9625000953674316,
|
| 561 |
+
"rewards/chosen": 7.245983123779297,
|
| 562 |
+
"rewards/margins": 34.91341018676758,
|
| 563 |
+
"rewards/rejected": -27.667430877685547,
|
| 564 |
+
"step": 370
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 2.5182724252491693,
|
| 568 |
+
"grad_norm": 0.47833314538002014,
|
| 569 |
+
"learning_rate": 4.650372869738415e-06,
|
| 570 |
+
"logits/chosen": -0.17948493361473083,
|
| 571 |
+
"logits/rejected": -0.2594072222709656,
|
| 572 |
+
"logps/chosen": -964.8648071289062,
|
| 573 |
+
"logps/rejected": -1580.651123046875,
|
| 574 |
+
"loss": 0.0309,
|
| 575 |
+
"rewards/accuracies": 0.957812488079071,
|
| 576 |
+
"rewards/chosen": 7.238793849945068,
|
| 577 |
+
"rewards/margins": 33.64303207397461,
|
| 578 |
+
"rewards/rejected": -26.404239654541016,
|
| 579 |
+
"step": 380
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 2.584717607973422,
|
| 583 |
+
"grad_norm": 0.4996240437030792,
|
| 584 |
+
"learning_rate": 4.620120240391065e-06,
|
| 585 |
+
"logits/chosen": -0.19269897043704987,
|
| 586 |
+
"logits/rejected": -0.25276264548301697,
|
| 587 |
+
"logps/chosen": -1004.8121948242188,
|
| 588 |
+
"logps/rejected": -1599.7081298828125,
|
| 589 |
+
"loss": 0.0263,
|
| 590 |
+
"rewards/accuracies": 0.9640625715255737,
|
| 591 |
+
"rewards/chosen": 7.900957107543945,
|
| 592 |
+
"rewards/margins": 33.70750427246094,
|
| 593 |
+
"rewards/rejected": -25.806547164916992,
|
| 594 |
+
"step": 390
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 2.6511627906976747,
|
| 598 |
+
"grad_norm": 0.15727798640727997,
|
| 599 |
+
"learning_rate": 4.588719528532342e-06,
|
| 600 |
+
"logits/chosen": -0.1740064173936844,
|
| 601 |
+
"logits/rejected": -0.25867804884910583,
|
| 602 |
+
"logps/chosen": -946.6286010742188,
|
| 603 |
+
"logps/rejected": -1478.2218017578125,
|
| 604 |
+
"loss": 0.0325,
|
| 605 |
+
"rewards/accuracies": 0.9578125476837158,
|
| 606 |
+
"rewards/chosen": 8.154074668884277,
|
| 607 |
+
"rewards/margins": 30.49738311767578,
|
| 608 |
+
"rewards/rejected": -22.343307495117188,
|
| 609 |
+
"step": 400
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 2.717607973421927,
|
| 613 |
+
"grad_norm": 0.02712377905845642,
|
| 614 |
+
"learning_rate": 4.556187738201656e-06,
|
| 615 |
+
"logits/chosen": -0.19258640706539154,
|
| 616 |
+
"logits/rejected": -0.2623384892940521,
|
| 617 |
+
"logps/chosen": -974.1800537109375,
|
| 618 |
+
"logps/rejected": -1545.2799072265625,
|
| 619 |
+
"loss": 0.0369,
|
| 620 |
+
"rewards/accuracies": 0.949999988079071,
|
| 621 |
+
"rewards/chosen": 5.771888732910156,
|
| 622 |
+
"rewards/margins": 35.3669548034668,
|
| 623 |
+
"rewards/rejected": -29.595062255859375,
|
| 624 |
+
"step": 410
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 2.7840531561461797,
|
| 628 |
+
"grad_norm": 0.4458315968513489,
|
| 629 |
+
"learning_rate": 4.522542485937369e-06,
|
| 630 |
+
"logits/chosen": -0.1969211995601654,
|
| 631 |
+
"logits/rejected": -0.2592199444770813,
|
| 632 |
+
"logps/chosen": -996.4677124023438,
|
| 633 |
+
"logps/rejected": -1540.347900390625,
|
| 634 |
+
"loss": 0.0307,
|
| 635 |
+
"rewards/accuracies": 0.9578126072883606,
|
| 636 |
+
"rewards/chosen": 8.071084022521973,
|
| 637 |
+
"rewards/margins": 33.89667892456055,
|
| 638 |
+
"rewards/rejected": -25.82559585571289,
|
| 639 |
+
"step": 420
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 2.850498338870432,
|
| 643 |
+
"grad_norm": 0.0796787217259407,
|
| 644 |
+
"learning_rate": 4.48780199123712e-06,
|
| 645 |
+
"logits/chosen": -0.19366417825222015,
|
| 646 |
+
"logits/rejected": -0.25863632559776306,
|
| 647 |
+
"logps/chosen": -1018.47607421875,
|
| 648 |
+
"logps/rejected": -1551.6494140625,
|
| 649 |
+
"loss": 0.0326,
|
| 650 |
+
"rewards/accuracies": 0.9546875357627869,
|
| 651 |
+
"rewards/chosen": 8.464451789855957,
|
| 652 |
+
"rewards/margins": 33.95619201660156,
|
| 653 |
+
"rewards/rejected": -25.491741180419922,
|
| 654 |
+
"step": 430
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 2.9169435215946846,
|
| 658 |
+
"grad_norm": 0.07867090404033661,
|
| 659 |
+
"learning_rate": 4.451985066691649e-06,
|
| 660 |
+
"logits/chosen": -0.19828085601329803,
|
| 661 |
+
"logits/rejected": -0.263224720954895,
|
| 662 |
+
"logps/chosen": -987.7084350585938,
|
| 663 |
+
"logps/rejected": -1536.230224609375,
|
| 664 |
+
"loss": 0.0312,
|
| 665 |
+
"rewards/accuracies": 0.956250011920929,
|
| 666 |
+
"rewards/chosen": 7.850325584411621,
|
| 667 |
+
"rewards/margins": 34.385406494140625,
|
| 668 |
+
"rewards/rejected": -26.535079956054688,
|
| 669 |
+
"step": 440
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.983388704318937,
|
| 673 |
+
"grad_norm": 0.07914525270462036,
|
| 674 |
+
"learning_rate": 4.415111107797445e-06,
|
| 675 |
+
"logits/chosen": -0.18529188632965088,
|
| 676 |
+
"logits/rejected": -0.25437110662460327,
|
| 677 |
+
"logps/chosen": -1014.2196044921875,
|
| 678 |
+
"logps/rejected": -1524.4237060546875,
|
| 679 |
+
"loss": 0.0269,
|
| 680 |
+
"rewards/accuracies": 0.9656250476837158,
|
| 681 |
+
"rewards/chosen": 7.967530727386475,
|
| 682 |
+
"rewards/margins": 34.628875732421875,
|
| 683 |
+
"rewards/rejected": -26.661344528198242,
|
| 684 |
+
"step": 450
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 3.046511627906977,
|
| 688 |
+
"grad_norm": 0.026503894478082657,
|
| 689 |
+
"learning_rate": 4.377200082453748e-06,
|
| 690 |
+
"logits/chosen": -0.18043489754199982,
|
| 691 |
+
"logits/rejected": -0.2650010883808136,
|
| 692 |
+
"logps/chosen": -929.5118408203125,
|
| 693 |
+
"logps/rejected": -1481.138427734375,
|
| 694 |
+
"loss": 0.0311,
|
| 695 |
+
"rewards/accuracies": 0.9539474248886108,
|
| 696 |
+
"rewards/chosen": 7.918570041656494,
|
| 697 |
+
"rewards/margins": 33.1231575012207,
|
| 698 |
+
"rewards/rejected": -25.204586029052734,
|
| 699 |
+
"step": 460
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 3.1129568106312293,
|
| 703 |
+
"grad_norm": 0.055876098573207855,
|
| 704 |
+
"learning_rate": 4.338272520149572e-06,
|
| 705 |
+
"logits/chosen": -0.20002666115760803,
|
| 706 |
+
"logits/rejected": -0.27018633484840393,
|
| 707 |
+
"logps/chosen": -1009.8369750976562,
|
| 708 |
+
"logps/rejected": -1604.3740234375,
|
| 709 |
+
"loss": 0.033,
|
| 710 |
+
"rewards/accuracies": 0.953125,
|
| 711 |
+
"rewards/chosen": 7.928904056549072,
|
| 712 |
+
"rewards/margins": 36.35823440551758,
|
| 713 |
+
"rewards/rejected": -28.4293270111084,
|
| 714 |
+
"step": 470
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 3.179401993355482,
|
| 718 |
+
"grad_norm": 0.007625865284353495,
|
| 719 |
+
"learning_rate": 4.2983495008466285e-06,
|
| 720 |
+
"logits/chosen": -0.1756007969379425,
|
| 721 |
+
"logits/rejected": -0.2624135911464691,
|
| 722 |
+
"logps/chosen": -979.9953002929688,
|
| 723 |
+
"logps/rejected": -1648.5302734375,
|
| 724 |
+
"loss": 0.027,
|
| 725 |
+
"rewards/accuracies": 0.9625000357627869,
|
| 726 |
+
"rewards/chosen": 6.49752950668335,
|
| 727 |
+
"rewards/margins": 39.52600860595703,
|
| 728 |
+
"rewards/rejected": -33.028480529785156,
|
| 729 |
+
"step": 480
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 3.2458471760797343,
|
| 733 |
+
"grad_norm": 1.3655072450637817,
|
| 734 |
+
"learning_rate": 4.257452643564155e-06,
|
| 735 |
+
"logits/chosen": -0.19411802291870117,
|
| 736 |
+
"logits/rejected": -0.26652225852012634,
|
| 737 |
+
"logps/chosen": -955.16552734375,
|
| 738 |
+
"logps/rejected": -1548.7850341796875,
|
| 739 |
+
"loss": 0.0344,
|
| 740 |
+
"rewards/accuracies": 0.953125,
|
| 741 |
+
"rewards/chosen": 7.868605613708496,
|
| 742 |
+
"rewards/margins": 34.91292190551758,
|
| 743 |
+
"rewards/rejected": -27.04431915283203,
|
| 744 |
+
"step": 490
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 3.3122923588039868,
|
| 748 |
+
"grad_norm": 0.03723715990781784,
|
| 749 |
+
"learning_rate": 4.215604094671835e-06,
|
| 750 |
+
"logits/chosen": -0.20745989680290222,
|
| 751 |
+
"logits/rejected": -0.26310446858406067,
|
| 752 |
+
"logps/chosen": -1038.3160400390625,
|
| 753 |
+
"logps/rejected": -1586.4140625,
|
| 754 |
+
"loss": 0.0265,
|
| 755 |
+
"rewards/accuracies": 0.9625000357627869,
|
| 756 |
+
"rewards/chosen": 8.868291854858398,
|
| 757 |
+
"rewards/margins": 36.5751953125,
|
| 758 |
+
"rewards/rejected": -27.70690155029297,
|
| 759 |
+
"step": 500
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 3.3787375415282392,
|
| 763 |
+
"grad_norm": 0.04950110614299774,
|
| 764 |
+
"learning_rate": 4.172826515897146e-06,
|
| 765 |
+
"logits/chosen": -0.19769521057605743,
|
| 766 |
+
"logits/rejected": -0.2655687928199768,
|
| 767 |
+
"logps/chosen": -1002.1111450195312,
|
| 768 |
+
"logps/rejected": -1597.6588134765625,
|
| 769 |
+
"loss": 0.0232,
|
| 770 |
+
"rewards/accuracies": 0.9671874642372131,
|
| 771 |
+
"rewards/chosen": 7.961508274078369,
|
| 772 |
+
"rewards/margins": 36.75872039794922,
|
| 773 |
+
"rewards/rejected": -28.79721450805664,
|
| 774 |
+
"step": 510
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 3.4451827242524917,
|
| 778 |
+
"grad_norm": 0.27452749013900757,
|
| 779 |
+
"learning_rate": 4.129143072053639e-06,
|
| 780 |
+
"logits/chosen": -0.18965086340904236,
|
| 781 |
+
"logits/rejected": -0.2614585757255554,
|
| 782 |
+
"logps/chosen": -1033.2958984375,
|
| 783 |
+
"logps/rejected": -1618.4097900390625,
|
| 784 |
+
"loss": 0.0222,
|
| 785 |
+
"rewards/accuracies": 0.96875,
|
| 786 |
+
"rewards/chosen": 8.251802444458008,
|
| 787 |
+
"rewards/margins": 37.35887145996094,
|
| 788 |
+
"rewards/rejected": -29.107072830200195,
|
| 789 |
+
"step": 520
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 3.511627906976744,
|
| 793 |
+
"grad_norm": 0.0644262358546257,
|
| 794 |
+
"learning_rate": 4.084577418496775e-06,
|
| 795 |
+
"logits/chosen": -0.20613916218280792,
|
| 796 |
+
"logits/rejected": -0.27108556032180786,
|
| 797 |
+
"logps/chosen": -1004.0643310546875,
|
| 798 |
+
"logps/rejected": -1601.927978515625,
|
| 799 |
+
"loss": 0.0287,
|
| 800 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 801 |
+
"rewards/chosen": 8.70193099975586,
|
| 802 |
+
"rewards/margins": 35.765296936035156,
|
| 803 |
+
"rewards/rejected": -27.063365936279297,
|
| 804 |
+
"step": 530
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.5780730897009967,
|
| 808 |
+
"grad_norm": 0.021257763728499413,
|
| 809 |
+
"learning_rate": 4.039153688314146e-06,
|
| 810 |
+
"logits/chosen": -0.20685432851314545,
|
| 811 |
+
"logits/rejected": -0.267654150724411,
|
| 812 |
+
"logps/chosen": -997.7388305664062,
|
| 813 |
+
"logps/rejected": -1588.1624755859375,
|
| 814 |
+
"loss": 0.0328,
|
| 815 |
+
"rewards/accuracies": 0.953125,
|
| 816 |
+
"rewards/chosen": 8.601181983947754,
|
| 817 |
+
"rewards/margins": 37.05242156982422,
|
| 818 |
+
"rewards/rejected": -28.45123863220215,
|
| 819 |
+
"step": 540
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 3.644518272425249,
|
| 823 |
+
"grad_norm": 0.029319366440176964,
|
| 824 |
+
"learning_rate": 3.992896479256966e-06,
|
| 825 |
+
"logits/chosen": -0.20922623574733734,
|
| 826 |
+
"logits/rejected": -0.2706463634967804,
|
| 827 |
+
"logps/chosen": -986.989501953125,
|
| 828 |
+
"logps/rejected": -1553.56787109375,
|
| 829 |
+
"loss": 0.0361,
|
| 830 |
+
"rewards/accuracies": 0.9484374523162842,
|
| 831 |
+
"rewards/chosen": 7.84389591217041,
|
| 832 |
+
"rewards/margins": 38.18416213989258,
|
| 833 |
+
"rewards/rejected": -30.34027099609375,
|
| 834 |
+
"step": 550
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 3.7109634551495017,
|
| 838 |
+
"grad_norm": 0.029560599476099014,
|
| 839 |
+
"learning_rate": 3.945830840419966e-06,
|
| 840 |
+
"logits/chosen": -0.18561449646949768,
|
| 841 |
+
"logits/rejected": -0.2662014961242676,
|
| 842 |
+
"logps/chosen": -1032.16552734375,
|
| 843 |
+
"logps/rejected": -1648.7718505859375,
|
| 844 |
+
"loss": 0.0263,
|
| 845 |
+
"rewards/accuracies": 0.9625000357627869,
|
| 846 |
+
"rewards/chosen": 8.428018569946289,
|
| 847 |
+
"rewards/margins": 39.801116943359375,
|
| 848 |
+
"rewards/rejected": -31.373104095458984,
|
| 849 |
+
"step": 560
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 3.777408637873754,
|
| 853 |
+
"grad_norm": 0.010398728772997856,
|
| 854 |
+
"learning_rate": 3.897982258676867e-06,
|
| 855 |
+
"logits/chosen": -0.1940029114484787,
|
| 856 |
+
"logits/rejected": -0.27436965703964233,
|
| 857 |
+
"logps/chosen": -961.5516357421875,
|
| 858 |
+
"logps/rejected": -1579.232421875,
|
| 859 |
+
"loss": 0.0383,
|
| 860 |
+
"rewards/accuracies": 0.9453125,
|
| 861 |
+
"rewards/chosen": 8.185991287231445,
|
| 862 |
+
"rewards/margins": 38.32309341430664,
|
| 863 |
+
"rewards/rejected": -30.13710594177246,
|
| 864 |
+
"step": 570
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 3.8438538205980066,
|
| 868 |
+
"grad_norm": 0.05186082422733307,
|
| 869 |
+
"learning_rate": 3.849376644878783e-06,
|
| 870 |
+
"logits/chosen": -0.20401492714881897,
|
| 871 |
+
"logits/rejected": -0.266576886177063,
|
| 872 |
+
"logps/chosen": -1019.6111450195312,
|
| 873 |
+
"logps/rejected": -1620.943115234375,
|
| 874 |
+
"loss": 0.0279,
|
| 875 |
+
"rewards/accuracies": 0.9609374403953552,
|
| 876 |
+
"rewards/chosen": 7.710155487060547,
|
| 877 |
+
"rewards/margins": 40.542640686035156,
|
| 878 |
+
"rewards/rejected": -32.83248519897461,
|
| 879 |
+
"step": 580
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 3.910299003322259,
|
| 883 |
+
"grad_norm": 0.11145040392875671,
|
| 884 |
+
"learning_rate": 3.8000403198230385e-06,
|
| 885 |
+
"logits/chosen": -0.19729003310203552,
|
| 886 |
+
"logits/rejected": -0.26546692848205566,
|
| 887 |
+
"logps/chosen": -987.747802734375,
|
| 888 |
+
"logps/rejected": -1537.516357421875,
|
| 889 |
+
"loss": 0.0223,
|
| 890 |
+
"rewards/accuracies": 0.9703125357627869,
|
| 891 |
+
"rewards/chosen": 8.000187873840332,
|
| 892 |
+
"rewards/margins": 37.64095687866211,
|
| 893 |
+
"rewards/rejected": -29.640769958496094,
|
| 894 |
+
"step": 590
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 3.9767441860465116,
|
| 898 |
+
"grad_norm": 0.2694355249404907,
|
| 899 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 900 |
+
"logits/chosen": -0.21104276180267334,
|
| 901 |
+
"logits/rejected": -0.2754373848438263,
|
| 902 |
+
"logps/chosen": -982.5044555664062,
|
| 903 |
+
"logps/rejected": -1553.8619384765625,
|
| 904 |
+
"loss": 0.0359,
|
| 905 |
+
"rewards/accuracies": 0.949999988079071,
|
| 906 |
+
"rewards/chosen": 8.334994316101074,
|
| 907 |
+
"rewards/margins": 38.43461227416992,
|
| 908 |
+
"rewards/rejected": -30.099620819091797,
|
| 909 |
+
"step": 600
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 4.039867109634551,
|
| 913 |
+
"grad_norm": 0.0788518562912941,
|
| 914 |
+
"learning_rate": 3.699282783125616e-06,
|
| 915 |
+
"logits/chosen": -0.20658275485038757,
|
| 916 |
+
"logits/rejected": -0.2766464054584503,
|
| 917 |
+
"logps/chosen": -965.9616088867188,
|
| 918 |
+
"logps/rejected": -1610.8004150390625,
|
| 919 |
+
"loss": 0.0344,
|
| 920 |
+
"rewards/accuracies": 0.9490132331848145,
|
| 921 |
+
"rewards/chosen": 5.881260395050049,
|
| 922 |
+
"rewards/margins": 41.5827522277832,
|
| 923 |
+
"rewards/rejected": -35.70149230957031,
|
| 924 |
+
"step": 610
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 4.106312292358804,
|
| 928 |
+
"grad_norm": 0.005860505159944296,
|
| 929 |
+
"learning_rate": 3.6479161334675294e-06,
|
| 930 |
+
"logits/chosen": -0.18060919642448425,
|
| 931 |
+
"logits/rejected": -0.2699647545814514,
|
| 932 |
+
"logps/chosen": -948.4402465820312,
|
| 933 |
+
"logps/rejected": -1546.55078125,
|
| 934 |
+
"loss": 0.02,
|
| 935 |
+
"rewards/accuracies": 0.971875011920929,
|
| 936 |
+
"rewards/chosen": 9.497621536254883,
|
| 937 |
+
"rewards/margins": 34.928138732910156,
|
| 938 |
+
"rewards/rejected": -25.430519104003906,
|
| 939 |
+
"step": 620
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 4.172757475083056,
|
| 943 |
+
"grad_norm": 0.006049708928912878,
|
| 944 |
+
"learning_rate": 3.595927866972694e-06,
|
| 945 |
+
"logits/chosen": -0.20163992047309875,
|
| 946 |
+
"logits/rejected": -0.2756229043006897,
|
| 947 |
+
"logps/chosen": -980.5719604492188,
|
| 948 |
+
"logps/rejected": -1553.8719482421875,
|
| 949 |
+
"loss": 0.0253,
|
| 950 |
+
"rewards/accuracies": 0.964062511920929,
|
| 951 |
+
"rewards/chosen": 9.457831382751465,
|
| 952 |
+
"rewards/margins": 36.089698791503906,
|
| 953 |
+
"rewards/rejected": -26.631864547729492,
|
| 954 |
+
"step": 630
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 4.239202657807309,
|
| 958 |
+
"grad_norm": 0.035230789333581924,
|
| 959 |
+
"learning_rate": 3.543346136204545e-06,
|
| 960 |
+
"logits/chosen": -0.20396091043949127,
|
| 961 |
+
"logits/rejected": -0.2794302701950073,
|
| 962 |
+
"logps/chosen": -1021.5272827148438,
|
| 963 |
+
"logps/rejected": -1694.7557373046875,
|
| 964 |
+
"loss": 0.0339,
|
| 965 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 966 |
+
"rewards/chosen": 8.177035331726074,
|
| 967 |
+
"rewards/margins": 42.25918197631836,
|
| 968 |
+
"rewards/rejected": -34.08214569091797,
|
| 969 |
+
"step": 640
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 4.305647840531561,
|
| 973 |
+
"grad_norm": 0.10100215673446655,
|
| 974 |
+
"learning_rate": 3.4901994150978926e-06,
|
| 975 |
+
"logits/chosen": -0.22341015934944153,
|
| 976 |
+
"logits/rejected": -0.27249380946159363,
|
| 977 |
+
"logps/chosen": -1055.6053466796875,
|
| 978 |
+
"logps/rejected": -1636.560302734375,
|
| 979 |
+
"loss": 0.0264,
|
| 980 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 981 |
+
"rewards/chosen": 7.556117057800293,
|
| 982 |
+
"rewards/margins": 41.31330490112305,
|
| 983 |
+
"rewards/rejected": -33.75718688964844,
|
| 984 |
+
"step": 650
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 4.372093023255814,
|
| 988 |
+
"grad_norm": 0.023471888154745102,
|
| 989 |
+
"learning_rate": 3.436516483539781e-06,
|
| 990 |
+
"logits/chosen": -0.21736814081668854,
|
| 991 |
+
"logits/rejected": -0.2824287712574005,
|
| 992 |
+
"logps/chosen": -982.9537353515625,
|
| 993 |
+
"logps/rejected": -1562.0203857421875,
|
| 994 |
+
"loss": 0.0416,
|
| 995 |
+
"rewards/accuracies": 0.940625011920929,
|
| 996 |
+
"rewards/chosen": 6.172419548034668,
|
| 997 |
+
"rewards/margins": 38.94424819946289,
|
| 998 |
+
"rewards/rejected": -32.771827697753906,
|
| 999 |
+
"step": 660
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 4.438538205980066,
|
| 1003 |
+
"grad_norm": 0.6410449743270874,
|
| 1004 |
+
"learning_rate": 3.3823264117846722e-06,
|
| 1005 |
+
"logits/chosen": -0.20028334856033325,
|
| 1006 |
+
"logits/rejected": -0.27208542823791504,
|
| 1007 |
+
"logps/chosen": -1042.419189453125,
|
| 1008 |
+
"logps/rejected": -1699.8818359375,
|
| 1009 |
+
"loss": 0.0243,
|
| 1010 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1011 |
+
"rewards/chosen": 6.371281623840332,
|
| 1012 |
+
"rewards/margins": 44.25002670288086,
|
| 1013 |
+
"rewards/rejected": -37.878746032714844,
|
| 1014 |
+
"step": 670
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 4.504983388704319,
|
| 1018 |
+
"grad_norm": 0.052079688757658005,
|
| 1019 |
+
"learning_rate": 3.3276585447123957e-06,
|
| 1020 |
+
"logits/chosen": -0.19798581302165985,
|
| 1021 |
+
"logits/rejected": -0.27162185311317444,
|
| 1022 |
+
"logps/chosen": -993.3763427734375,
|
| 1023 |
+
"logps/rejected": -1604.8309326171875,
|
| 1024 |
+
"loss": 0.0232,
|
| 1025 |
+
"rewards/accuracies": 0.9671875238418579,
|
| 1026 |
+
"rewards/chosen": 8.744821548461914,
|
| 1027 |
+
"rewards/margins": 39.64046859741211,
|
| 1028 |
+
"rewards/rejected": -30.895647048950195,
|
| 1029 |
+
"step": 680
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 4.571428571428571,
|
| 1033 |
+
"grad_norm": 0.010166758671402931,
|
| 1034 |
+
"learning_rate": 3.272542485937369e-06,
|
| 1035 |
+
"logits/chosen": -0.2120884358882904,
|
| 1036 |
+
"logits/rejected": -0.2828371226787567,
|
| 1037 |
+
"logps/chosen": -979.7803344726562,
|
| 1038 |
+
"logps/rejected": -1523.5648193359375,
|
| 1039 |
+
"loss": 0.0295,
|
| 1040 |
+
"rewards/accuracies": 0.9578125476837158,
|
| 1041 |
+
"rewards/chosen": 9.28367805480957,
|
| 1042 |
+
"rewards/margins": 36.00359344482422,
|
| 1043 |
+
"rewards/rejected": -26.71991729736328,
|
| 1044 |
+
"step": 690
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 4.637873754152824,
|
| 1048 |
+
"grad_norm": 0.07122723013162613,
|
| 1049 |
+
"learning_rate": 3.217008081777726e-06,
|
| 1050 |
+
"logits/chosen": -0.2192113846540451,
|
| 1051 |
+
"logits/rejected": -0.2890257239341736,
|
| 1052 |
+
"logps/chosen": -929.1891479492188,
|
| 1053 |
+
"logps/rejected": -1475.6949462890625,
|
| 1054 |
+
"loss": 0.0346,
|
| 1055 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 1056 |
+
"rewards/chosen": 9.672334671020508,
|
| 1057 |
+
"rewards/margins": 35.26881408691406,
|
| 1058 |
+
"rewards/rejected": -25.596477508544922,
|
| 1059 |
+
"step": 700
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 4.704318936877076,
|
| 1063 |
+
"grad_norm": 0.005518330726772547,
|
| 1064 |
+
"learning_rate": 3.1610854050930063e-06,
|
| 1065 |
+
"logits/chosen": -0.19891753792762756,
|
| 1066 |
+
"logits/rejected": -0.27730634808540344,
|
| 1067 |
+
"logps/chosen": -985.259033203125,
|
| 1068 |
+
"logps/rejected": -1633.97314453125,
|
| 1069 |
+
"loss": 0.0294,
|
| 1070 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1071 |
+
"rewards/chosen": 9.867067337036133,
|
| 1072 |
+
"rewards/margins": 38.44176483154297,
|
| 1073 |
+
"rewards/rejected": -28.574697494506836,
|
| 1074 |
+
"step": 710
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 4.770764119601329,
|
| 1078 |
+
"grad_norm": 0.008760541677474976,
|
| 1079 |
+
"learning_rate": 3.1048047389991693e-06,
|
| 1080 |
+
"logits/chosen": -0.21026046574115753,
|
| 1081 |
+
"logits/rejected": -0.2759222686290741,
|
| 1082 |
+
"logps/chosen": -1001.1782836914062,
|
| 1083 |
+
"logps/rejected": -1598.6923828125,
|
| 1084 |
+
"loss": 0.0328,
|
| 1085 |
+
"rewards/accuracies": 0.9546875357627869,
|
| 1086 |
+
"rewards/chosen": 9.485105514526367,
|
| 1087 |
+
"rewards/margins": 38.465606689453125,
|
| 1088 |
+
"rewards/rejected": -28.980497360229492,
|
| 1089 |
+
"step": 720
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 4.837209302325581,
|
| 1093 |
+
"grad_norm": 0.016581548377871513,
|
| 1094 |
+
"learning_rate": 3.0481965604697582e-06,
|
| 1095 |
+
"logits/chosen": -0.1935136914253235,
|
| 1096 |
+
"logits/rejected": -0.27545979619026184,
|
| 1097 |
+
"logps/chosen": -973.7349243164062,
|
| 1098 |
+
"logps/rejected": -1601.532958984375,
|
| 1099 |
+
"loss": 0.0242,
|
| 1100 |
+
"rewards/accuracies": 0.9656250476837158,
|
| 1101 |
+
"rewards/chosen": 7.491199016571045,
|
| 1102 |
+
"rewards/margins": 41.66270065307617,
|
| 1103 |
+
"rewards/rejected": -34.1714973449707,
|
| 1104 |
+
"step": 730
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 4.903654485049834,
|
| 1108 |
+
"grad_norm": 0.00849311426281929,
|
| 1109 |
+
"learning_rate": 2.9912915238320755e-06,
|
| 1110 |
+
"logits/chosen": -0.19976982474327087,
|
| 1111 |
+
"logits/rejected": -0.2815185487270355,
|
| 1112 |
+
"logps/chosen": -965.1561889648438,
|
| 1113 |
+
"logps/rejected": -1615.7647705078125,
|
| 1114 |
+
"loss": 0.0306,
|
| 1115 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1116 |
+
"rewards/chosen": 9.341843605041504,
|
| 1117 |
+
"rewards/margins": 40.90797424316406,
|
| 1118 |
+
"rewards/rejected": -31.566131591796875,
|
| 1119 |
+
"step": 740
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 4.970099667774086,
|
| 1123 |
+
"grad_norm": 0.018370211124420166,
|
| 1124 |
+
"learning_rate": 2.9341204441673267e-06,
|
| 1125 |
+
"logits/chosen": -0.21805454790592194,
|
| 1126 |
+
"logits/rejected": -0.28051772713661194,
|
| 1127 |
+
"logps/chosen": -1029.127197265625,
|
| 1128 |
+
"logps/rejected": -1687.1690673828125,
|
| 1129 |
+
"loss": 0.0305,
|
| 1130 |
+
"rewards/accuracies": 0.9562500715255737,
|
| 1131 |
+
"rewards/chosen": 9.458334922790527,
|
| 1132 |
+
"rewards/margins": 42.93328857421875,
|
| 1133 |
+
"rewards/rejected": -33.47495651245117,
|
| 1134 |
+
"step": 750
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 5.033222591362127,
|
| 1138 |
+
"grad_norm": 0.014480801299214363,
|
| 1139 |
+
"learning_rate": 2.876714280623708e-06,
|
| 1140 |
+
"logits/chosen": -0.19559527933597565,
|
| 1141 |
+
"logits/rejected": -0.2727430462837219,
|
| 1142 |
+
"logps/chosen": -966.6165771484375,
|
| 1143 |
+
"logps/rejected": -1623.1119384765625,
|
| 1144 |
+
"loss": 0.023,
|
| 1145 |
+
"rewards/accuracies": 0.9654605388641357,
|
| 1146 |
+
"rewards/chosen": 8.72402572631836,
|
| 1147 |
+
"rewards/margins": 41.55656051635742,
|
| 1148 |
+
"rewards/rejected": -32.83253479003906,
|
| 1149 |
+
"step": 760
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 5.099667774086379,
|
| 1153 |
+
"grad_norm": 0.004520288668572903,
|
| 1154 |
+
"learning_rate": 2.8191041196514874e-06,
|
| 1155 |
+
"logits/chosen": -0.19683074951171875,
|
| 1156 |
+
"logits/rejected": -0.28631192445755005,
|
| 1157 |
+
"logps/chosen": -1009.6803588867188,
|
| 1158 |
+
"logps/rejected": -1692.927734375,
|
| 1159 |
+
"loss": 0.0359,
|
| 1160 |
+
"rewards/accuracies": 0.9484374523162842,
|
| 1161 |
+
"rewards/chosen": 8.311098098754883,
|
| 1162 |
+
"rewards/margins": 43.74188232421875,
|
| 1163 |
+
"rewards/rejected": -35.430789947509766,
|
| 1164 |
+
"step": 770
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 5.166112956810632,
|
| 1168 |
+
"grad_norm": 0.03138303384184837,
|
| 1169 |
+
"learning_rate": 2.761321158169134e-06,
|
| 1170 |
+
"logits/chosen": -0.21386271715164185,
|
| 1171 |
+
"logits/rejected": -0.27936917543411255,
|
| 1172 |
+
"logps/chosen": -1063.0811767578125,
|
| 1173 |
+
"logps/rejected": -1719.7852783203125,
|
| 1174 |
+
"loss": 0.0338,
|
| 1175 |
+
"rewards/accuracies": 0.953125,
|
| 1176 |
+
"rewards/chosen": 8.667400360107422,
|
| 1177 |
+
"rewards/margins": 45.44544219970703,
|
| 1178 |
+
"rewards/rejected": -36.77804183959961,
|
| 1179 |
+
"step": 780
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 5.232558139534884,
|
| 1183 |
+
"grad_norm": 0.05668414384126663,
|
| 1184 |
+
"learning_rate": 2.703396686669646e-06,
|
| 1185 |
+
"logits/chosen": -0.2076747715473175,
|
| 1186 |
+
"logits/rejected": -0.2838803827762604,
|
| 1187 |
+
"logps/chosen": -986.5599365234375,
|
| 1188 |
+
"logps/rejected": -1629.951416015625,
|
| 1189 |
+
"loss": 0.0339,
|
| 1190 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 1191 |
+
"rewards/chosen": 8.366827011108398,
|
| 1192 |
+
"rewards/margins": 41.922176361083984,
|
| 1193 |
+
"rewards/rejected": -33.55535125732422,
|
| 1194 |
+
"step": 790
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 5.299003322259137,
|
| 1198 |
+
"grad_norm": 0.02641725167632103,
|
| 1199 |
+
"learning_rate": 2.6453620722761897e-06,
|
| 1200 |
+
"logits/chosen": -0.21938087046146393,
|
| 1201 |
+
"logits/rejected": -0.27865853905677795,
|
| 1202 |
+
"logps/chosen": -1019.5682373046875,
|
| 1203 |
+
"logps/rejected": -1662.2581787109375,
|
| 1204 |
+
"loss": 0.0305,
|
| 1205 |
+
"rewards/accuracies": 0.9562499523162842,
|
| 1206 |
+
"rewards/chosen": 8.933462142944336,
|
| 1207 |
+
"rewards/margins": 44.147613525390625,
|
| 1208 |
+
"rewards/rejected": -35.21416091918945,
|
| 1209 |
+
"step": 800
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 5.365448504983389,
|
| 1213 |
+
"grad_norm": 0.02291504666209221,
|
| 1214 |
+
"learning_rate": 2.587248741756253e-06,
|
| 1215 |
+
"logits/chosen": -0.19530998170375824,
|
| 1216 |
+
"logits/rejected": -0.2803831100463867,
|
| 1217 |
+
"logps/chosen": -1002.9342651367188,
|
| 1218 |
+
"logps/rejected": -1661.538330078125,
|
| 1219 |
+
"loss": 0.0202,
|
| 1220 |
+
"rewards/accuracies": 0.9718750715255737,
|
| 1221 |
+
"rewards/chosen": 7.737525939941406,
|
| 1222 |
+
"rewards/margins": 43.79603576660156,
|
| 1223 |
+
"rewards/rejected": -36.058509826660156,
|
| 1224 |
+
"step": 810
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 5.431893687707642,
|
| 1228 |
+
"grad_norm": 0.008839444257318974,
|
| 1229 |
+
"learning_rate": 2.5290881645034932e-06,
|
| 1230 |
+
"logits/chosen": -0.20212310552597046,
|
| 1231 |
+
"logits/rejected": -0.26888853311538696,
|
| 1232 |
+
"logps/chosen": -986.2118530273438,
|
| 1233 |
+
"logps/rejected": -1612.09619140625,
|
| 1234 |
+
"loss": 0.0196,
|
| 1235 |
+
"rewards/accuracies": 0.9734375476837158,
|
| 1236 |
+
"rewards/chosen": 7.7221360206604,
|
| 1237 |
+
"rewards/margins": 42.720664978027344,
|
| 1238 |
+
"rewards/rejected": -34.998531341552734,
|
| 1239 |
+
"step": 820
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 5.498338870431894,
|
| 1243 |
+
"grad_norm": 0.01382855698466301,
|
| 1244 |
+
"learning_rate": 2.470911835496508e-06,
|
| 1245 |
+
"logits/chosen": -0.19646425545215607,
|
| 1246 |
+
"logits/rejected": -0.2737523019313812,
|
| 1247 |
+
"logps/chosen": -1009.5712890625,
|
| 1248 |
+
"logps/rejected": -1698.4268798828125,
|
| 1249 |
+
"loss": 0.024,
|
| 1250 |
+
"rewards/accuracies": 0.9656250476837158,
|
| 1251 |
+
"rewards/chosen": 7.582469940185547,
|
| 1252 |
+
"rewards/margins": 45.48985290527344,
|
| 1253 |
+
"rewards/rejected": -37.90738296508789,
|
| 1254 |
+
"step": 830
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 5.5647840531561465,
|
| 1258 |
+
"grad_norm": 0.22888414561748505,
|
| 1259 |
+
"learning_rate": 2.4127512582437486e-06,
|
| 1260 |
+
"logits/chosen": -0.21698302030563354,
|
| 1261 |
+
"logits/rejected": -0.28401094675064087,
|
| 1262 |
+
"logps/chosen": -955.4769897460938,
|
| 1263 |
+
"logps/rejected": -1580.6790771484375,
|
| 1264 |
+
"loss": 0.0351,
|
| 1265 |
+
"rewards/accuracies": 0.9500000476837158,
|
| 1266 |
+
"rewards/chosen": 8.103876113891602,
|
| 1267 |
+
"rewards/margins": 41.412132263183594,
|
| 1268 |
+
"rewards/rejected": -33.308258056640625,
|
| 1269 |
+
"step": 840
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 5.631229235880399,
|
| 1273 |
+
"grad_norm": 0.024031994864344597,
|
| 1274 |
+
"learning_rate": 2.3546379277238107e-06,
|
| 1275 |
+
"logits/chosen": -0.2063540369272232,
|
| 1276 |
+
"logits/rejected": -0.2823134660720825,
|
| 1277 |
+
"logps/chosen": -986.14453125,
|
| 1278 |
+
"logps/rejected": -1621.3382568359375,
|
| 1279 |
+
"loss": 0.023,
|
| 1280 |
+
"rewards/accuracies": 0.9671875238418579,
|
| 1281 |
+
"rewards/chosen": 9.615578651428223,
|
| 1282 |
+
"rewards/margins": 43.22168731689453,
|
| 1283 |
+
"rewards/rejected": -33.60611343383789,
|
| 1284 |
+
"step": 850
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 5.6976744186046515,
|
| 1288 |
+
"grad_norm": 0.010551623068749905,
|
| 1289 |
+
"learning_rate": 2.296603313330355e-06,
|
| 1290 |
+
"logits/chosen": -0.22434450685977936,
|
| 1291 |
+
"logits/rejected": -0.28748488426208496,
|
| 1292 |
+
"logps/chosen": -952.8927612304688,
|
| 1293 |
+
"logps/rejected": -1560.9144287109375,
|
| 1294 |
+
"loss": 0.0393,
|
| 1295 |
+
"rewards/accuracies": 0.9437499642372131,
|
| 1296 |
+
"rewards/chosen": 9.026927947998047,
|
| 1297 |
+
"rewards/margins": 40.4544563293457,
|
| 1298 |
+
"rewards/rejected": -31.42752456665039,
|
| 1299 |
+
"step": 860
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 5.764119601328904,
|
| 1303 |
+
"grad_norm": 0.0801558867096901,
|
| 1304 |
+
"learning_rate": 2.238678841830867e-06,
|
| 1305 |
+
"logits/chosen": -0.21255464851856232,
|
| 1306 |
+
"logits/rejected": -0.28065019845962524,
|
| 1307 |
+
"logps/chosen": -972.204345703125,
|
| 1308 |
+
"logps/rejected": -1601.0872802734375,
|
| 1309 |
+
"loss": 0.0326,
|
| 1310 |
+
"rewards/accuracies": 0.9531249403953552,
|
| 1311 |
+
"rewards/chosen": 8.012575149536133,
|
| 1312 |
+
"rewards/margins": 41.84084701538086,
|
| 1313 |
+
"rewards/rejected": -33.828269958496094,
|
| 1314 |
+
"step": 870
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 5.8305647840531565,
|
| 1318 |
+
"grad_norm": 0.08310822397470474,
|
| 1319 |
+
"learning_rate": 2.1808958803485134e-06,
|
| 1320 |
+
"logits/chosen": -0.20110653340816498,
|
| 1321 |
+
"logits/rejected": -0.28692275285720825,
|
| 1322 |
+
"logps/chosen": -959.0699462890625,
|
| 1323 |
+
"logps/rejected": -1659.7845458984375,
|
| 1324 |
+
"loss": 0.0294,
|
| 1325 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1326 |
+
"rewards/chosen": 8.209933280944824,
|
| 1327 |
+
"rewards/margins": 44.37035369873047,
|
| 1328 |
+
"rewards/rejected": -36.160423278808594,
|
| 1329 |
+
"step": 880
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 5.897009966777409,
|
| 1333 |
+
"grad_norm": 0.012136918492615223,
|
| 1334 |
+
"learning_rate": 2.1232857193762923e-06,
|
| 1335 |
+
"logits/chosen": -0.20550034940242767,
|
| 1336 |
+
"logits/rejected": -0.2773197889328003,
|
| 1337 |
+
"logps/chosen": -976.6769409179688,
|
| 1338 |
+
"logps/rejected": -1591.7425537109375,
|
| 1339 |
+
"loss": 0.0284,
|
| 1340 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 1341 |
+
"rewards/chosen": 8.299397468566895,
|
| 1342 |
+
"rewards/margins": 42.518890380859375,
|
| 1343 |
+
"rewards/rejected": -34.21949005126953,
|
| 1344 |
+
"step": 890
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 5.9634551495016614,
|
| 1348 |
+
"grad_norm": 0.10679075866937637,
|
| 1349 |
+
"learning_rate": 2.0658795558326745e-06,
|
| 1350 |
+
"logits/chosen": -0.21557196974754333,
|
| 1351 |
+
"logits/rejected": -0.2806122899055481,
|
| 1352 |
+
"logps/chosen": -991.53466796875,
|
| 1353 |
+
"logps/rejected": -1629.1478271484375,
|
| 1354 |
+
"loss": 0.0316,
|
| 1355 |
+
"rewards/accuracies": 0.9546875357627869,
|
| 1356 |
+
"rewards/chosen": 8.746665000915527,
|
| 1357 |
+
"rewards/margins": 42.621334075927734,
|
| 1358 |
+
"rewards/rejected": -33.87466812133789,
|
| 1359 |
+
"step": 900
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 6.026578073089701,
|
| 1363 |
+
"grad_norm": 0.016590170562267303,
|
| 1364 |
+
"learning_rate": 2.0087084761679245e-06,
|
| 1365 |
+
"logits/chosen": -0.20582108199596405,
|
| 1366 |
+
"logits/rejected": -0.27399370074272156,
|
| 1367 |
+
"logps/chosen": -1017.154541015625,
|
| 1368 |
+
"logps/rejected": -1689.7642822265625,
|
| 1369 |
+
"loss": 0.0176,
|
| 1370 |
+
"rewards/accuracies": 0.9736842513084412,
|
| 1371 |
+
"rewards/chosen": 8.807604789733887,
|
| 1372 |
+
"rewards/margins": 44.51702880859375,
|
| 1373 |
+
"rewards/rejected": -35.70942306518555,
|
| 1374 |
+
"step": 910
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 6.093023255813954,
|
| 1378 |
+
"grad_norm": 0.02025614120066166,
|
| 1379 |
+
"learning_rate": 1.9518034395302413e-06,
|
| 1380 |
+
"logits/chosen": -0.19750240445137024,
|
| 1381 |
+
"logits/rejected": -0.2810463011264801,
|
| 1382 |
+
"logps/chosen": -990.0672607421875,
|
| 1383 |
+
"logps/rejected": -1701.806640625,
|
| 1384 |
+
"loss": 0.0273,
|
| 1385 |
+
"rewards/accuracies": 0.9609375596046448,
|
| 1386 |
+
"rewards/chosen": 9.167813301086426,
|
| 1387 |
+
"rewards/margins": 45.0367546081543,
|
| 1388 |
+
"rewards/rejected": -35.86894226074219,
|
| 1389 |
+
"step": 920
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 6.159468438538206,
|
| 1393 |
+
"grad_norm": 0.08283823728561401,
|
| 1394 |
+
"learning_rate": 1.895195261000831e-06,
|
| 1395 |
+
"logits/chosen": -0.20525017380714417,
|
| 1396 |
+
"logits/rejected": -0.2784285545349121,
|
| 1397 |
+
"logps/chosen": -1004.66357421875,
|
| 1398 |
+
"logps/rejected": -1686.80908203125,
|
| 1399 |
+
"loss": 0.0261,
|
| 1400 |
+
"rewards/accuracies": 0.9625000357627869,
|
| 1401 |
+
"rewards/chosen": 8.760817527770996,
|
| 1402 |
+
"rewards/margins": 44.76042556762695,
|
| 1403 |
+
"rewards/rejected": -35.99960708618164,
|
| 1404 |
+
"step": 930
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 6.225913621262459,
|
| 1408 |
+
"grad_norm": 0.03424767032265663,
|
| 1409 |
+
"learning_rate": 1.8389145949069953e-06,
|
| 1410 |
+
"logits/chosen": -0.20580098032951355,
|
| 1411 |
+
"logits/rejected": -0.2847464382648468,
|
| 1412 |
+
"logps/chosen": -1016.2109985351562,
|
| 1413 |
+
"logps/rejected": -1675.8634033203125,
|
| 1414 |
+
"loss": 0.0298,
|
| 1415 |
+
"rewards/accuracies": 0.9578125476837158,
|
| 1416 |
+
"rewards/chosen": 8.987253189086914,
|
| 1417 |
+
"rewards/margins": 45.16158676147461,
|
| 1418 |
+
"rewards/rejected": -36.174339294433594,
|
| 1419 |
+
"step": 940
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 6.292358803986711,
|
| 1423 |
+
"grad_norm": 0.08232512325048447,
|
| 1424 |
+
"learning_rate": 1.7829919182222752e-06,
|
| 1425 |
+
"logits/chosen": -0.21305060386657715,
|
| 1426 |
+
"logits/rejected": -0.278297483921051,
|
| 1427 |
+
"logps/chosen": -994.125,
|
| 1428 |
+
"logps/rejected": -1603.1475830078125,
|
| 1429 |
+
"loss": 0.0284,
|
| 1430 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 1431 |
+
"rewards/chosen": 8.597787857055664,
|
| 1432 |
+
"rewards/margins": 43.382965087890625,
|
| 1433 |
+
"rewards/rejected": -34.785179138183594,
|
| 1434 |
+
"step": 950
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 6.358803986710964,
|
| 1438 |
+
"grad_norm": 0.11056870222091675,
|
| 1439 |
+
"learning_rate": 1.7274575140626318e-06,
|
| 1440 |
+
"logits/chosen": -0.20215469598770142,
|
| 1441 |
+
"logits/rejected": -0.2870931327342987,
|
| 1442 |
+
"logps/chosen": -979.541748046875,
|
| 1443 |
+
"logps/rejected": -1649.0330810546875,
|
| 1444 |
+
"loss": 0.0327,
|
| 1445 |
+
"rewards/accuracies": 0.9546874761581421,
|
| 1446 |
+
"rewards/chosen": 8.763008117675781,
|
| 1447 |
+
"rewards/margins": 43.981746673583984,
|
| 1448 |
+
"rewards/rejected": -35.21874237060547,
|
| 1449 |
+
"step": 960
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 6.425249169435216,
|
| 1453 |
+
"grad_norm": 0.17400754988193512,
|
| 1454 |
+
"learning_rate": 1.6723414552876052e-06,
|
| 1455 |
+
"logits/chosen": -0.2209278792142868,
|
| 1456 |
+
"logits/rejected": -0.29013967514038086,
|
| 1457 |
+
"logps/chosen": -1003.218505859375,
|
| 1458 |
+
"logps/rejected": -1650.18798828125,
|
| 1459 |
+
"loss": 0.0316,
|
| 1460 |
+
"rewards/accuracies": 0.9546874761581421,
|
| 1461 |
+
"rewards/chosen": 9.620824813842773,
|
| 1462 |
+
"rewards/margins": 42.64148712158203,
|
| 1463 |
+
"rewards/rejected": -33.020660400390625,
|
| 1464 |
+
"step": 970
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 6.4916943521594686,
|
| 1468 |
+
"grad_norm": 0.007690763566643,
|
| 1469 |
+
"learning_rate": 1.6176735882153284e-06,
|
| 1470 |
+
"logits/chosen": -0.2107953131198883,
|
| 1471 |
+
"logits/rejected": -0.28245460987091064,
|
| 1472 |
+
"logps/chosen": -986.3475341796875,
|
| 1473 |
+
"logps/rejected": -1621.1788330078125,
|
| 1474 |
+
"loss": 0.0274,
|
| 1475 |
+
"rewards/accuracies": 0.9609375,
|
| 1476 |
+
"rewards/chosen": 9.415980339050293,
|
| 1477 |
+
"rewards/margins": 43.30890655517578,
|
| 1478 |
+
"rewards/rejected": -33.89292526245117,
|
| 1479 |
+
"step": 980
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 6.558139534883721,
|
| 1483 |
+
"grad_norm": 0.007917268201708794,
|
| 1484 |
+
"learning_rate": 1.56348351646022e-06,
|
| 1485 |
+
"logits/chosen": -0.21639874577522278,
|
| 1486 |
+
"logits/rejected": -0.28619349002838135,
|
| 1487 |
+
"logps/chosen": -996.1339111328125,
|
| 1488 |
+
"logps/rejected": -1701.57275390625,
|
| 1489 |
+
"loss": 0.0326,
|
| 1490 |
+
"rewards/accuracies": 0.953125,
|
| 1491 |
+
"rewards/chosen": 8.72823715209961,
|
| 1492 |
+
"rewards/margins": 45.77238082885742,
|
| 1493 |
+
"rewards/rejected": -37.04414367675781,
|
| 1494 |
+
"step": 990
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 6.6245847176079735,
|
| 1498 |
+
"grad_norm": 0.009819849394261837,
|
| 1499 |
+
"learning_rate": 1.509800584902108e-06,
|
| 1500 |
+
"logits/chosen": -0.20866619050502777,
|
| 1501 |
+
"logits/rejected": -0.2797132730484009,
|
| 1502 |
+
"logps/chosen": -966.6695556640625,
|
| 1503 |
+
"logps/rejected": -1632.124267578125,
|
| 1504 |
+
"loss": 0.025,
|
| 1505 |
+
"rewards/accuracies": 0.9640624523162842,
|
| 1506 |
+
"rewards/chosen": 8.59074878692627,
|
| 1507 |
+
"rewards/margins": 44.0283317565918,
|
| 1508 |
+
"rewards/rejected": -35.437583923339844,
|
| 1509 |
+
"step": 1000
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 6.691029900332226,
|
| 1513 |
+
"grad_norm": 0.004776698537170887,
|
| 1514 |
+
"learning_rate": 1.4566538637954556e-06,
|
| 1515 |
+
"logits/chosen": -0.20434224605560303,
|
| 1516 |
+
"logits/rejected": -0.2861843407154083,
|
| 1517 |
+
"logps/chosen": -919.8916625976562,
|
| 1518 |
+
"logps/rejected": -1539.8953857421875,
|
| 1519 |
+
"loss": 0.0229,
|
| 1520 |
+
"rewards/accuracies": 0.9671874642372131,
|
| 1521 |
+
"rewards/chosen": 8.691563606262207,
|
| 1522 |
+
"rewards/margins": 40.7747917175293,
|
| 1523 |
+
"rewards/rejected": -32.083229064941406,
|
| 1524 |
+
"step": 1010
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 6.7574750830564785,
|
| 1528 |
+
"grad_norm": 0.00468341913074255,
|
| 1529 |
+
"learning_rate": 1.4040721330273063e-06,
|
| 1530 |
+
"logits/chosen": -0.22420468926429749,
|
| 1531 |
+
"logits/rejected": -0.28962579369544983,
|
| 1532 |
+
"logps/chosen": -963.7784423828125,
|
| 1533 |
+
"logps/rejected": -1591.2889404296875,
|
| 1534 |
+
"loss": 0.0434,
|
| 1535 |
+
"rewards/accuracies": 0.9375,
|
| 1536 |
+
"rewards/chosen": 8.361918449401855,
|
| 1537 |
+
"rewards/margins": 43.0437126159668,
|
| 1538 |
+
"rewards/rejected": -34.68179702758789,
|
| 1539 |
+
"step": 1020
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 6.823920265780731,
|
| 1543 |
+
"grad_norm": 0.04172283038496971,
|
| 1544 |
+
"learning_rate": 1.3520838665324704e-06,
|
| 1545 |
+
"logits/chosen": -0.20492787659168243,
|
| 1546 |
+
"logits/rejected": -0.2808915078639984,
|
| 1547 |
+
"logps/chosen": -975.69921875,
|
| 1548 |
+
"logps/rejected": -1621.455322265625,
|
| 1549 |
+
"loss": 0.0295,
|
| 1550 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1551 |
+
"rewards/chosen": 8.498809814453125,
|
| 1552 |
+
"rewards/margins": 43.53432846069336,
|
| 1553 |
+
"rewards/rejected": -35.035518646240234,
|
| 1554 |
+
"step": 1030
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 6.8903654485049834,
|
| 1558 |
+
"grad_norm": 0.014652963727712631,
|
| 1559 |
+
"learning_rate": 1.3007172168743854e-06,
|
| 1560 |
+
"logits/chosen": -0.2149275690317154,
|
| 1561 |
+
"logits/rejected": -0.28474923968315125,
|
| 1562 |
+
"logps/chosen": -996.0318603515625,
|
| 1563 |
+
"logps/rejected": -1641.5811767578125,
|
| 1564 |
+
"loss": 0.0273,
|
| 1565 |
+
"rewards/accuracies": 0.9609375,
|
| 1566 |
+
"rewards/chosen": 9.519927024841309,
|
| 1567 |
+
"rewards/margins": 42.85387420654297,
|
| 1568 |
+
"rewards/rejected": -33.33395004272461,
|
| 1569 |
+
"step": 1040
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 6.956810631229236,
|
| 1573 |
+
"grad_norm": 0.058657530695199966,
|
| 1574 |
+
"learning_rate": 1.2500000000000007e-06,
|
| 1575 |
+
"logits/chosen": -0.19909979403018951,
|
| 1576 |
+
"logits/rejected": -0.2798536419868469,
|
| 1577 |
+
"logps/chosen": -1014.5499877929688,
|
| 1578 |
+
"logps/rejected": -1707.7406005859375,
|
| 1579 |
+
"loss": 0.024,
|
| 1580 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1581 |
+
"rewards/chosen": 9.100263595581055,
|
| 1582 |
+
"rewards/margins": 45.508270263671875,
|
| 1583 |
+
"rewards/rejected": -36.40800857543945,
|
| 1584 |
+
"step": 1050
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 7.019933554817276,
|
| 1588 |
+
"grad_norm": 0.08728719502687454,
|
| 1589 |
+
"learning_rate": 1.1999596801769617e-06,
|
| 1590 |
+
"logits/chosen": -0.23542040586471558,
|
| 1591 |
+
"logits/rejected": -0.28797173500061035,
|
| 1592 |
+
"logps/chosen": -1009.5910034179688,
|
| 1593 |
+
"logps/rejected": -1648.842529296875,
|
| 1594 |
+
"loss": 0.0337,
|
| 1595 |
+
"rewards/accuracies": 0.9490132331848145,
|
| 1596 |
+
"rewards/chosen": 9.306107521057129,
|
| 1597 |
+
"rewards/margins": 43.96856689453125,
|
| 1598 |
+
"rewards/rejected": -34.66246032714844,
|
| 1599 |
+
"step": 1060
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 7.086378737541528,
|
| 1603 |
+
"grad_norm": 0.02003307454288006,
|
| 1604 |
+
"learning_rate": 1.1506233551212186e-06,
|
| 1605 |
+
"logits/chosen": -0.20915324985980988,
|
| 1606 |
+
"logits/rejected": -0.2867971360683441,
|
| 1607 |
+
"logps/chosen": -1014.2936401367188,
|
| 1608 |
+
"logps/rejected": -1682.6341552734375,
|
| 1609 |
+
"loss": 0.024,
|
| 1610 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1611 |
+
"rewards/chosen": 9.585081100463867,
|
| 1612 |
+
"rewards/margins": 44.993507385253906,
|
| 1613 |
+
"rewards/rejected": -35.408424377441406,
|
| 1614 |
+
"step": 1070
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 7.152823920265781,
|
| 1618 |
+
"grad_norm": 0.006420216057449579,
|
| 1619 |
+
"learning_rate": 1.1020177413231334e-06,
|
| 1620 |
+
"logits/chosen": -0.20691779255867004,
|
| 1621 |
+
"logits/rejected": -0.2857610583305359,
|
| 1622 |
+
"logps/chosen": -972.8970947265625,
|
| 1623 |
+
"logps/rejected": -1629.0289306640625,
|
| 1624 |
+
"loss": 0.0304,
|
| 1625 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1626 |
+
"rewards/chosen": 8.977264404296875,
|
| 1627 |
+
"rewards/margins": 43.42811584472656,
|
| 1628 |
+
"rewards/rejected": -34.45084762573242,
|
| 1629 |
+
"step": 1080
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 7.219269102990033,
|
| 1633 |
+
"grad_norm": 0.02167355827987194,
|
| 1634 |
+
"learning_rate": 1.0541691595800338e-06,
|
| 1635 |
+
"logits/chosen": -0.20276153087615967,
|
| 1636 |
+
"logits/rejected": -0.2794567942619324,
|
| 1637 |
+
"logps/chosen": -1003.0272216796875,
|
| 1638 |
+
"logps/rejected": -1687.5604248046875,
|
| 1639 |
+
"loss": 0.0348,
|
| 1640 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1641 |
+
"rewards/chosen": 8.437175750732422,
|
| 1642 |
+
"rewards/margins": 46.04411315917969,
|
| 1643 |
+
"rewards/rejected": -37.60693359375,
|
| 1644 |
+
"step": 1090
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 7.285714285714286,
|
| 1648 |
+
"grad_norm": 0.05616578087210655,
|
| 1649 |
+
"learning_rate": 1.0071035207430352e-06,
|
| 1650 |
+
"logits/chosen": -0.19716766476631165,
|
| 1651 |
+
"logits/rejected": -0.28074419498443604,
|
| 1652 |
+
"logps/chosen": -981.2333374023438,
|
| 1653 |
+
"logps/rejected": -1657.1065673828125,
|
| 1654 |
+
"loss": 0.0262,
|
| 1655 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1656 |
+
"rewards/chosen": 8.663593292236328,
|
| 1657 |
+
"rewards/margins": 45.19432067871094,
|
| 1658 |
+
"rewards/rejected": -36.530723571777344,
|
| 1659 |
+
"step": 1100
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 7.352159468438538,
|
| 1663 |
+
"grad_norm": 0.01851947233080864,
|
| 1664 |
+
"learning_rate": 9.608463116858544e-07,
|
| 1665 |
+
"logits/chosen": -0.2091461569070816,
|
| 1666 |
+
"logits/rejected": -0.2789131700992584,
|
| 1667 |
+
"logps/chosen": -1033.9593505859375,
|
| 1668 |
+
"logps/rejected": -1729.9864501953125,
|
| 1669 |
+
"loss": 0.0217,
|
| 1670 |
+
"rewards/accuracies": 0.96875,
|
| 1671 |
+
"rewards/chosen": 9.394617080688477,
|
| 1672 |
+
"rewards/margins": 47.00334930419922,
|
| 1673 |
+
"rewards/rejected": -37.608734130859375,
|
| 1674 |
+
"step": 1110
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 7.4186046511627906,
|
| 1678 |
+
"grad_norm": 0.005292957648634911,
|
| 1679 |
+
"learning_rate": 9.154225815032242e-07,
|
| 1680 |
+
"logits/chosen": -0.2048484981060028,
|
| 1681 |
+
"logits/rejected": -0.28349366784095764,
|
| 1682 |
+
"logps/chosen": -944.3374633789062,
|
| 1683 |
+
"logps/rejected": -1572.23681640625,
|
| 1684 |
+
"loss": 0.0284,
|
| 1685 |
+
"rewards/accuracies": 0.9593749642372131,
|
| 1686 |
+
"rewards/chosen": 9.14864444732666,
|
| 1687 |
+
"rewards/margins": 42.550880432128906,
|
| 1688 |
+
"rewards/rejected": -33.40223693847656,
|
| 1689 |
+
"step": 1120
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 7.485049833887043,
|
| 1693 |
+
"grad_norm": 0.026964331045746803,
|
| 1694 |
+
"learning_rate": 8.708569279463622e-07,
|
| 1695 |
+
"logits/chosen": -0.20627908408641815,
|
| 1696 |
+
"logits/rejected": -0.2784706652164459,
|
| 1697 |
+
"logps/chosen": -964.9845581054688,
|
| 1698 |
+
"logps/rejected": -1633.6651611328125,
|
| 1699 |
+
"loss": 0.0273,
|
| 1700 |
+
"rewards/accuracies": 0.9609375,
|
| 1701 |
+
"rewards/chosen": 8.720355987548828,
|
| 1702 |
+
"rewards/margins": 44.15267562866211,
|
| 1703 |
+
"rewards/rejected": -35.43231964111328,
|
| 1704 |
+
"step": 1130
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 7.5514950166112955,
|
| 1708 |
+
"grad_norm": 0.0011573478113859892,
|
| 1709 |
+
"learning_rate": 8.271734841028553e-07,
|
| 1710 |
+
"logits/chosen": -0.21377238631248474,
|
| 1711 |
+
"logits/rejected": -0.2910877466201782,
|
| 1712 |
+
"logps/chosen": -972.2564086914062,
|
| 1713 |
+
"logps/rejected": -1676.005126953125,
|
| 1714 |
+
"loss": 0.0305,
|
| 1715 |
+
"rewards/accuracies": 0.9578125476837158,
|
| 1716 |
+
"rewards/chosen": 9.1951322555542,
|
| 1717 |
+
"rewards/margins": 45.21895980834961,
|
| 1718 |
+
"rewards/rejected": -36.02383041381836,
|
| 1719 |
+
"step": 1140
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 7.617940199335548,
|
| 1723 |
+
"grad_norm": 0.028651030734181404,
|
| 1724 |
+
"learning_rate": 7.843959053281663e-07,
|
| 1725 |
+
"logits/chosen": -0.217359721660614,
|
| 1726 |
+
"logits/rejected": -0.2867337465286255,
|
| 1727 |
+
"logps/chosen": -986.8521118164062,
|
| 1728 |
+
"logps/rejected": -1608.333740234375,
|
| 1729 |
+
"loss": 0.024,
|
| 1730 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1731 |
+
"rewards/chosen": 8.933571815490723,
|
| 1732 |
+
"rewards/margins": 43.58317947387695,
|
| 1733 |
+
"rewards/rejected": -34.64960861206055,
|
| 1734 |
+
"step": 1150
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 7.6843853820598005,
|
| 1738 |
+
"grad_norm": 0.021432682871818542,
|
| 1739 |
+
"learning_rate": 7.425473564358457e-07,
|
| 1740 |
+
"logits/chosen": -0.21459349989891052,
|
| 1741 |
+
"logits/rejected": -0.2918223738670349,
|
| 1742 |
+
"logps/chosen": -950.2572021484375,
|
| 1743 |
+
"logps/rejected": -1665.85546875,
|
| 1744 |
+
"loss": 0.0348,
|
| 1745 |
+
"rewards/accuracies": 0.9499999284744263,
|
| 1746 |
+
"rewards/chosen": 8.907698631286621,
|
| 1747 |
+
"rewards/margins": 44.95459747314453,
|
| 1748 |
+
"rewards/rejected": -36.04690170288086,
|
| 1749 |
+
"step": 1160
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 7.750830564784053,
|
| 1753 |
+
"grad_norm": 0.0003203708620276302,
|
| 1754 |
+
"learning_rate": 7.016504991533727e-07,
|
| 1755 |
+
"logits/chosen": -0.2105288803577423,
|
| 1756 |
+
"logits/rejected": -0.2897786796092987,
|
| 1757 |
+
"logps/chosen": -1022.5786743164062,
|
| 1758 |
+
"logps/rejected": -1684.5330810546875,
|
| 1759 |
+
"loss": 0.0273,
|
| 1760 |
+
"rewards/accuracies": 0.9609375,
|
| 1761 |
+
"rewards/chosen": 8.936477661132812,
|
| 1762 |
+
"rewards/margins": 44.89566421508789,
|
| 1763 |
+
"rewards/rejected": -35.95918655395508,
|
| 1764 |
+
"step": 1170
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 7.8172757475083055,
|
| 1768 |
+
"grad_norm": 0.04623947665095329,
|
| 1769 |
+
"learning_rate": 6.617274798504286e-07,
|
| 1770 |
+
"logits/chosen": -0.22387103736400604,
|
| 1771 |
+
"logits/rejected": -0.28619539737701416,
|
| 1772 |
+
"logps/chosen": -971.675048828125,
|
| 1773 |
+
"logps/rejected": -1651.590087890625,
|
| 1774 |
+
"loss": 0.038,
|
| 1775 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 1776 |
+
"rewards/chosen": 8.927178382873535,
|
| 1777 |
+
"rewards/margins": 44.58891677856445,
|
| 1778 |
+
"rewards/rejected": -35.661739349365234,
|
| 1779 |
+
"step": 1180
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 7.883720930232558,
|
| 1783 |
+
"grad_norm": 0.1079607829451561,
|
| 1784 |
+
"learning_rate": 6.227999175462521e-07,
|
| 1785 |
+
"logits/chosen": -0.2090545892715454,
|
| 1786 |
+
"logits/rejected": -0.2881450355052948,
|
| 1787 |
+
"logps/chosen": -952.6052856445312,
|
| 1788 |
+
"logps/rejected": -1569.903076171875,
|
| 1789 |
+
"loss": 0.03,
|
| 1790 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1791 |
+
"rewards/chosen": 9.027518272399902,
|
| 1792 |
+
"rewards/margins": 41.79720687866211,
|
| 1793 |
+
"rewards/rejected": -32.76968765258789,
|
| 1794 |
+
"step": 1190
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 7.95016611295681,
|
| 1798 |
+
"grad_norm": 0.0075605567544698715,
|
| 1799 |
+
"learning_rate": 5.848888922025553e-07,
|
| 1800 |
+
"logits/chosen": -0.21150845289230347,
|
| 1801 |
+
"logits/rejected": -0.28404462337493896,
|
| 1802 |
+
"logps/chosen": -996.9921875,
|
| 1803 |
+
"logps/rejected": -1636.0484619140625,
|
| 1804 |
+
"loss": 0.0282,
|
| 1805 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 1806 |
+
"rewards/chosen": 9.349788665771484,
|
| 1807 |
+
"rewards/margins": 43.83129119873047,
|
| 1808 |
+
"rewards/rejected": -34.481502532958984,
|
| 1809 |
+
"step": 1200
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 8.013289036544851,
|
| 1813 |
+
"grad_norm": 0.10174239426851273,
|
| 1814 |
+
"learning_rate": 5.48014933308352e-07,
|
| 1815 |
+
"logits/chosen": -0.23533087968826294,
|
| 1816 |
+
"logits/rejected": -0.28401005268096924,
|
| 1817 |
+
"logps/chosen": -979.27587890625,
|
| 1818 |
+
"logps/rejected": -1561.8616943359375,
|
| 1819 |
+
"loss": 0.0337,
|
| 1820 |
+
"rewards/accuracies": 0.9490132331848145,
|
| 1821 |
+
"rewards/chosen": 9.154387474060059,
|
| 1822 |
+
"rewards/margins": 42.322410583496094,
|
| 1823 |
+
"rewards/rejected": -33.16802215576172,
|
| 1824 |
+
"step": 1210
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 8.079734219269103,
|
| 1828 |
+
"grad_norm": 0.037876877933740616,
|
| 1829 |
+
"learning_rate": 5.121980087628802e-07,
|
| 1830 |
+
"logits/chosen": -0.21666212379932404,
|
| 1831 |
+
"logits/rejected": -0.28196656703948975,
|
| 1832 |
+
"logps/chosen": -1002.8568725585938,
|
| 1833 |
+
"logps/rejected": -1634.5693359375,
|
| 1834 |
+
"loss": 0.0304,
|
| 1835 |
+
"rewards/accuracies": 0.9562499523162842,
|
| 1836 |
+
"rewards/chosen": 9.121562957763672,
|
| 1837 |
+
"rewards/margins": 44.36361312866211,
|
| 1838 |
+
"rewards/rejected": -35.24205017089844,
|
| 1839 |
+
"step": 1220
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 8.146179401993356,
|
| 1843 |
+
"grad_norm": 0.10557155311107635,
|
| 1844 |
+
"learning_rate": 4.774575140626317e-07,
|
| 1845 |
+
"logits/chosen": -0.19356323778629303,
|
| 1846 |
+
"logits/rejected": -0.27849042415618896,
|
| 1847 |
+
"logps/chosen": -967.4945068359375,
|
| 1848 |
+
"logps/rejected": -1631.644775390625,
|
| 1849 |
+
"loss": 0.0229,
|
| 1850 |
+
"rewards/accuracies": 0.9671875238418579,
|
| 1851 |
+
"rewards/chosen": 8.805334091186523,
|
| 1852 |
+
"rewards/margins": 43.998016357421875,
|
| 1853 |
+
"rewards/rejected": -35.19268035888672,
|
| 1854 |
+
"step": 1230
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 8.212624584717608,
|
| 1858 |
+
"grad_norm": 0.022014953196048737,
|
| 1859 |
+
"learning_rate": 4.438122617983442e-07,
|
| 1860 |
+
"logits/chosen": -0.21807442605495453,
|
| 1861 |
+
"logits/rejected": -0.28545916080474854,
|
| 1862 |
+
"logps/chosen": -999.7384033203125,
|
| 1863 |
+
"logps/rejected": -1630.1734619140625,
|
| 1864 |
+
"loss": 0.0283,
|
| 1865 |
+
"rewards/accuracies": 0.9593749642372131,
|
| 1866 |
+
"rewards/chosen": 9.582186698913574,
|
| 1867 |
+
"rewards/margins": 44.091854095458984,
|
| 1868 |
+
"rewards/rejected": -34.509674072265625,
|
| 1869 |
+
"step": 1240
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 8.279069767441861,
|
| 1873 |
+
"grad_norm": 0.007186461240053177,
|
| 1874 |
+
"learning_rate": 4.1128047146765936e-07,
|
| 1875 |
+
"logits/chosen": -0.2181546837091446,
|
| 1876 |
+
"logits/rejected": -0.28165513277053833,
|
| 1877 |
+
"logps/chosen": -1004.7896118164062,
|
| 1878 |
+
"logps/rejected": -1676.617431640625,
|
| 1879 |
+
"loss": 0.0263,
|
| 1880 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1881 |
+
"rewards/chosen": 9.609552383422852,
|
| 1882 |
+
"rewards/margins": 45.283634185791016,
|
| 1883 |
+
"rewards/rejected": -35.67407989501953,
|
| 1884 |
+
"step": 1250
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 8.345514950166113,
|
| 1888 |
+
"grad_norm": 0.05021677538752556,
|
| 1889 |
+
"learning_rate": 3.798797596089351e-07,
|
| 1890 |
+
"logits/chosen": -0.2084813416004181,
|
| 1891 |
+
"logits/rejected": -0.2937774360179901,
|
| 1892 |
+
"logps/chosen": -988.1944580078125,
|
| 1893 |
+
"logps/rejected": -1683.8162841796875,
|
| 1894 |
+
"loss": 0.0294,
|
| 1895 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1896 |
+
"rewards/chosen": 8.989810943603516,
|
| 1897 |
+
"rewards/margins": 45.19856262207031,
|
| 1898 |
+
"rewards/rejected": -36.20874786376953,
|
| 1899 |
+
"step": 1260
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 8.411960132890366,
|
| 1903 |
+
"grad_norm": 0.025954890996217728,
|
| 1904 |
+
"learning_rate": 3.4962713026158697e-07,
|
| 1905 |
+
"logits/chosen": -0.20687171816825867,
|
| 1906 |
+
"logits/rejected": -0.28650417923927307,
|
| 1907 |
+
"logps/chosen": -1004.6787109375,
|
| 1908 |
+
"logps/rejected": -1670.76513671875,
|
| 1909 |
+
"loss": 0.0305,
|
| 1910 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1911 |
+
"rewards/chosen": 9.011610984802246,
|
| 1912 |
+
"rewards/margins": 44.895050048828125,
|
| 1913 |
+
"rewards/rejected": -35.88343811035156,
|
| 1914 |
+
"step": 1270
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 8.478405315614618,
|
| 1918 |
+
"grad_norm": 0.0554281584918499,
|
| 1919 |
+
"learning_rate": 3.2053896575809426e-07,
|
| 1920 |
+
"logits/chosen": -0.2132522314786911,
|
| 1921 |
+
"logits/rejected": -0.28568530082702637,
|
| 1922 |
+
"logps/chosen": -999.4458618164062,
|
| 1923 |
+
"logps/rejected": -1643.4739990234375,
|
| 1924 |
+
"loss": 0.0336,
|
| 1925 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 1926 |
+
"rewards/chosen": 8.973052978515625,
|
| 1927 |
+
"rewards/margins": 44.0505485534668,
|
| 1928 |
+
"rewards/rejected": -35.07749938964844,
|
| 1929 |
+
"step": 1280
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 8.544850498338871,
|
| 1933 |
+
"grad_norm": 0.022469861432909966,
|
| 1934 |
+
"learning_rate": 2.9263101785268253e-07,
|
| 1935 |
+
"logits/chosen": -0.20100554823875427,
|
| 1936 |
+
"logits/rejected": -0.2842293679714203,
|
| 1937 |
+
"logps/chosen": -990.7542724609375,
|
| 1938 |
+
"logps/rejected": -1718.5325927734375,
|
| 1939 |
+
"loss": 0.0317,
|
| 1940 |
+
"rewards/accuracies": 0.9546875357627869,
|
| 1941 |
+
"rewards/chosen": 9.08223819732666,
|
| 1942 |
+
"rewards/margins": 47.4109001159668,
|
| 1943 |
+
"rewards/rejected": -38.32866668701172,
|
| 1944 |
+
"step": 1290
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 8.611295681063122,
|
| 1948 |
+
"grad_norm": 0.026393350213766098,
|
| 1949 |
+
"learning_rate": 2.6591839919146963e-07,
|
| 1950 |
+
"logits/chosen": -0.226187601685524,
|
| 1951 |
+
"logits/rejected": -0.29070550203323364,
|
| 1952 |
+
"logps/chosen": -976.7939453125,
|
| 1953 |
+
"logps/rejected": -1631.8603515625,
|
| 1954 |
+
"loss": 0.0337,
|
| 1955 |
+
"rewards/accuracies": 0.9515624642372131,
|
| 1956 |
+
"rewards/chosen": 9.085509300231934,
|
| 1957 |
+
"rewards/margins": 43.998069763183594,
|
| 1958 |
+
"rewards/rejected": -34.91256332397461,
|
| 1959 |
+
"step": 1300
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 8.677740863787376,
|
| 1963 |
+
"grad_norm": 0.0665309801697731,
|
| 1964 |
+
"learning_rate": 2.404155751286988e-07,
|
| 1965 |
+
"logits/chosen": -0.19709895551204681,
|
| 1966 |
+
"logits/rejected": -0.2848767340183258,
|
| 1967 |
+
"logps/chosen": -960.6867065429688,
|
| 1968 |
+
"logps/rejected": -1658.4659423828125,
|
| 1969 |
+
"loss": 0.0217,
|
| 1970 |
+
"rewards/accuracies": 0.9687501192092896,
|
| 1971 |
+
"rewards/chosen": 8.968391418457031,
|
| 1972 |
+
"rewards/margins": 44.709556579589844,
|
| 1973 |
+
"rewards/rejected": -35.74116134643555,
|
| 1974 |
+
"step": 1310
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 8.744186046511627,
|
| 1978 |
+
"grad_norm": 0.03218339383602142,
|
| 1979 |
+
"learning_rate": 2.1613635589349756e-07,
|
| 1980 |
+
"logits/chosen": -0.213968887925148,
|
| 1981 |
+
"logits/rejected": -0.2814236581325531,
|
| 1982 |
+
"logps/chosen": -1001.9435424804688,
|
| 1983 |
+
"logps/rejected": -1669.0272216796875,
|
| 1984 |
+
"loss": 0.0255,
|
| 1985 |
+
"rewards/accuracies": 0.964062511920929,
|
| 1986 |
+
"rewards/chosen": 8.87741756439209,
|
| 1987 |
+
"rewards/margins": 45.792850494384766,
|
| 1988 |
+
"rewards/rejected": -36.915435791015625,
|
| 1989 |
+
"step": 1320
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 8.81063122923588,
|
| 1993 |
+
"grad_norm": 0.027269069105386734,
|
| 1994 |
+
"learning_rate": 1.9309388911139427e-07,
|
| 1995 |
+
"logits/chosen": -0.21916161477565765,
|
| 1996 |
+
"logits/rejected": -0.2914758026599884,
|
| 1997 |
+
"logps/chosen": -955.6594848632812,
|
| 1998 |
+
"logps/rejected": -1604.225341796875,
|
| 1999 |
+
"loss": 0.038,
|
| 2000 |
+
"rewards/accuracies": 0.9453125596046448,
|
| 2001 |
+
"rewards/chosen": 9.038484573364258,
|
| 2002 |
+
"rewards/margins": 44.25259017944336,
|
| 2003 |
+
"rewards/rejected": -35.21410369873047,
|
| 2004 |
+
"step": 1330
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 8.877076411960132,
|
| 2008 |
+
"grad_norm": 0.041460830718278885,
|
| 2009 |
+
"learning_rate": 1.713006526846439e-07,
|
| 2010 |
+
"logits/chosen": -0.20554713904857635,
|
| 2011 |
+
"logits/rejected": -0.2785477638244629,
|
| 2012 |
+
"logps/chosen": -990.9168090820312,
|
| 2013 |
+
"logps/rejected": -1690.2587890625,
|
| 2014 |
+
"loss": 0.0197,
|
| 2015 |
+
"rewards/accuracies": 0.9718750715255737,
|
| 2016 |
+
"rewards/chosen": 9.279547691345215,
|
| 2017 |
+
"rewards/margins": 46.73269271850586,
|
| 2018 |
+
"rewards/rejected": -37.453147888183594,
|
| 2019 |
+
"step": 1340
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 8.943521594684386,
|
| 2023 |
+
"grad_norm": 0.013856984674930573,
|
| 2024 |
+
"learning_rate": 1.507684480352292e-07,
|
| 2025 |
+
"logits/chosen": -0.21384954452514648,
|
| 2026 |
+
"logits/rejected": -0.29137367010116577,
|
| 2027 |
+
"logps/chosen": -947.7948608398438,
|
| 2028 |
+
"logps/rejected": -1607.2012939453125,
|
| 2029 |
+
"loss": 0.037,
|
| 2030 |
+
"rewards/accuracies": 0.948437511920929,
|
| 2031 |
+
"rewards/chosen": 8.737092971801758,
|
| 2032 |
+
"rewards/margins": 43.15813446044922,
|
| 2033 |
+
"rewards/rejected": -34.42103958129883,
|
| 2034 |
+
"step": 1350
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 9.006644518272426,
|
| 2038 |
+
"grad_norm": 0.01655706949532032,
|
| 2039 |
+
"learning_rate": 1.31508393714177e-07,
|
| 2040 |
+
"logits/chosen": -0.22794920206069946,
|
| 2041 |
+
"logits/rejected": -0.28400349617004395,
|
| 2042 |
+
"logps/chosen": -989.8770141601562,
|
| 2043 |
+
"logps/rejected": -1598.807373046875,
|
| 2044 |
+
"loss": 0.0293,
|
| 2045 |
+
"rewards/accuracies": 0.955592155456543,
|
| 2046 |
+
"rewards/chosen": 9.594341278076172,
|
| 2047 |
+
"rewards/margins": 43.53114318847656,
|
| 2048 |
+
"rewards/rejected": -33.936805725097656,
|
| 2049 |
+
"step": 1360
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 9.073089700996677,
|
| 2053 |
+
"grad_norm": 0.017982784658670425,
|
| 2054 |
+
"learning_rate": 1.1353091938067024e-07,
|
| 2055 |
+
"logits/chosen": -0.22623415291309357,
|
| 2056 |
+
"logits/rejected": -0.29436349868774414,
|
| 2057 |
+
"logps/chosen": -972.5148315429688,
|
| 2058 |
+
"logps/rejected": -1602.6168212890625,
|
| 2059 |
+
"loss": 0.0402,
|
| 2060 |
+
"rewards/accuracies": 0.942187488079071,
|
| 2061 |
+
"rewards/chosen": 9.074629783630371,
|
| 2062 |
+
"rewards/margins": 43.110740661621094,
|
| 2063 |
+
"rewards/rejected": -34.036109924316406,
|
| 2064 |
+
"step": 1370
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 9.13953488372093,
|
| 2068 |
+
"grad_norm": 0.004334774799644947,
|
| 2069 |
+
"learning_rate": 9.684576015420277e-08,
|
| 2070 |
+
"logits/chosen": -0.22325854003429413,
|
| 2071 |
+
"logits/rejected": -0.28321751952171326,
|
| 2072 |
+
"logps/chosen": -982.11572265625,
|
| 2073 |
+
"logps/rejected": -1638.993896484375,
|
| 2074 |
+
"loss": 0.0272,
|
| 2075 |
+
"rewards/accuracies": 0.9609375596046448,
|
| 2076 |
+
"rewards/chosen": 9.52658462524414,
|
| 2077 |
+
"rewards/margins": 44.85720443725586,
|
| 2078 |
+
"rewards/rejected": -35.33061599731445,
|
| 2079 |
+
"step": 1380
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 9.205980066445182,
|
| 2083 |
+
"grad_norm": 0.018007798120379448,
|
| 2084 |
+
"learning_rate": 8.146195134284052e-08,
|
| 2085 |
+
"logits/chosen": -0.20939311385154724,
|
| 2086 |
+
"logits/rejected": -0.28719571232795715,
|
| 2087 |
+
"logps/chosen": -977.8973999023438,
|
| 2088 |
+
"logps/rejected": -1680.6883544921875,
|
| 2089 |
+
"loss": 0.0261,
|
| 2090 |
+
"rewards/accuracies": 0.9625000953674316,
|
| 2091 |
+
"rewards/chosen": 9.223061561584473,
|
| 2092 |
+
"rewards/margins": 44.93825912475586,
|
| 2093 |
+
"rewards/rejected": -35.71520233154297,
|
| 2094 |
+
"step": 1390
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 9.272425249169435,
|
| 2098 |
+
"grad_norm": 0.11176791042089462,
|
| 2099 |
+
"learning_rate": 6.738782355044048e-08,
|
| 2100 |
+
"logits/chosen": -0.2103874534368515,
|
| 2101 |
+
"logits/rejected": -0.28649529814720154,
|
| 2102 |
+
"logps/chosen": -950.8549194335938,
|
| 2103 |
+
"logps/rejected": -1609.9903564453125,
|
| 2104 |
+
"loss": 0.0316,
|
| 2105 |
+
"rewards/accuracies": 0.9546874761581421,
|
| 2106 |
+
"rewards/chosen": 8.95008659362793,
|
| 2107 |
+
"rewards/margins": 43.54734802246094,
|
| 2108 |
+
"rewards/rejected": -34.597259521484375,
|
| 2109 |
+
"step": 1400
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 9.338870431893687,
|
| 2113 |
+
"grad_norm": 0.027401011437177658,
|
| 2114 |
+
"learning_rate": 5.463099816548578e-08,
|
| 2115 |
+
"logits/chosen": -0.21140608191490173,
|
| 2116 |
+
"logits/rejected": -0.27913111448287964,
|
| 2117 |
+
"logps/chosen": -1046.7650146484375,
|
| 2118 |
+
"logps/rejected": -1688.9732666015625,
|
| 2119 |
+
"loss": 0.0186,
|
| 2120 |
+
"rewards/accuracies": 0.973437488079071,
|
| 2121 |
+
"rewards/chosen": 9.671619415283203,
|
| 2122 |
+
"rewards/margins": 46.846065521240234,
|
| 2123 |
+
"rewards/rejected": -37.1744499206543,
|
| 2124 |
+
"step": 1410
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 9.40531561461794,
|
| 2128 |
+
"grad_norm": 0.0032386966049671173,
|
| 2129 |
+
"learning_rate": 4.319838323396691e-08,
|
| 2130 |
+
"logits/chosen": -0.20964354276657104,
|
| 2131 |
+
"logits/rejected": -0.28780320286750793,
|
| 2132 |
+
"logps/chosen": -936.599853515625,
|
| 2133 |
+
"logps/rejected": -1573.99267578125,
|
| 2134 |
+
"loss": 0.0295,
|
| 2135 |
+
"rewards/accuracies": 0.957812488079071,
|
| 2136 |
+
"rewards/chosen": 8.730453491210938,
|
| 2137 |
+
"rewards/margins": 43.16553497314453,
|
| 2138 |
+
"rewards/rejected": -34.43508529663086,
|
| 2139 |
+
"step": 1420
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 9.471760797342192,
|
| 2143 |
+
"grad_norm": 0.004229187499731779,
|
| 2144 |
+
"learning_rate": 3.309616971855195e-08,
|
| 2145 |
+
"logits/chosen": -0.22262541949748993,
|
| 2146 |
+
"logits/rejected": -0.2844969928264618,
|
| 2147 |
+
"logps/chosen": -1004.188232421875,
|
| 2148 |
+
"logps/rejected": -1635.885009765625,
|
| 2149 |
+
"loss": 0.0293,
|
| 2150 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 2151 |
+
"rewards/chosen": 9.532754898071289,
|
| 2152 |
+
"rewards/margins": 44.6163215637207,
|
| 2153 |
+
"rewards/rejected": -35.08356475830078,
|
| 2154 |
+
"step": 1430
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 9.538205980066445,
|
| 2158 |
+
"grad_norm": 0.004731557797640562,
|
| 2159 |
+
"learning_rate": 2.4329828146074096e-08,
|
| 2160 |
+
"logits/chosen": -0.2104283571243286,
|
| 2161 |
+
"logits/rejected": -0.29052725434303284,
|
| 2162 |
+
"logps/chosen": -969.8198852539062,
|
| 2163 |
+
"logps/rejected": -1657.4737548828125,
|
| 2164 |
+
"loss": 0.0349,
|
| 2165 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2166 |
+
"rewards/chosen": 8.997271537780762,
|
| 2167 |
+
"rewards/margins": 44.33870315551758,
|
| 2168 |
+
"rewards/rejected": -35.3414306640625,
|
| 2169 |
+
"step": 1440
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 9.604651162790697,
|
| 2173 |
+
"grad_norm": 0.04136363044381142,
|
| 2174 |
+
"learning_rate": 1.6904105645142443e-08,
|
| 2175 |
+
"logits/chosen": -0.2100651115179062,
|
| 2176 |
+
"logits/rejected": -0.2873626947402954,
|
| 2177 |
+
"logps/chosen": -951.37109375,
|
| 2178 |
+
"logps/rejected": -1638.732666015625,
|
| 2179 |
+
"loss": 0.0298,
|
| 2180 |
+
"rewards/accuracies": 0.957812488079071,
|
| 2181 |
+
"rewards/chosen": 8.915160179138184,
|
| 2182 |
+
"rewards/margins": 44.380455017089844,
|
| 2183 |
+
"rewards/rejected": -35.46529769897461,
|
| 2184 |
+
"step": 1450
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 9.67109634551495,
|
| 2188 |
+
"grad_norm": 0.02028900943696499,
|
| 2189 |
+
"learning_rate": 1.0823023375489128e-08,
|
| 2190 |
+
"logits/chosen": -0.20784109830856323,
|
| 2191 |
+
"logits/rejected": -0.2790369689464569,
|
| 2192 |
+
"logps/chosen": -1006.0265502929688,
|
| 2193 |
+
"logps/rejected": -1686.586181640625,
|
| 2194 |
+
"loss": 0.0207,
|
| 2195 |
+
"rewards/accuracies": 0.9703124761581421,
|
| 2196 |
+
"rewards/chosen": 9.536187171936035,
|
| 2197 |
+
"rewards/margins": 46.156455993652344,
|
| 2198 |
+
"rewards/rejected": -36.62027359008789,
|
| 2199 |
+
"step": 1460
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 9.737541528239202,
|
| 2203 |
+
"grad_norm": 0.0592665858566761,
|
| 2204 |
+
"learning_rate": 6.089874350439507e-09,
|
| 2205 |
+
"logits/chosen": -0.20697462558746338,
|
| 2206 |
+
"logits/rejected": -0.2866511642932892,
|
| 2207 |
+
"logps/chosen": -991.3517456054688,
|
| 2208 |
+
"logps/rejected": -1683.9049072265625,
|
| 2209 |
+
"loss": 0.0338,
|
| 2210 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 2211 |
+
"rewards/chosen": 8.75904369354248,
|
| 2212 |
+
"rewards/margins": 46.561065673828125,
|
| 2213 |
+
"rewards/rejected": -37.80202865600586,
|
| 2214 |
+
"step": 1470
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 9.803986710963455,
|
| 2218 |
+
"grad_norm": 0.10093259811401367,
|
| 2219 |
+
"learning_rate": 2.7072216536885855e-09,
|
| 2220 |
+
"logits/chosen": -0.21343420445919037,
|
| 2221 |
+
"logits/rejected": -0.29003995656967163,
|
| 2222 |
+
"logps/chosen": -978.60205078125,
|
| 2223 |
+
"logps/rejected": -1612.8367919921875,
|
| 2224 |
+
"loss": 0.0359,
|
| 2225 |
+
"rewards/accuracies": 0.9484375715255737,
|
| 2226 |
+
"rewards/chosen": 8.994418144226074,
|
| 2227 |
+
"rewards/margins": 43.18799591064453,
|
| 2228 |
+
"rewards/rejected": -34.193572998046875,
|
| 2229 |
+
"step": 1480
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 9.870431893687707,
|
| 2233 |
+
"grad_norm": 0.0005512971547432244,
|
| 2234 |
+
"learning_rate": 6.768970513457151e-10,
|
| 2235 |
+
"logits/chosen": -0.20138706266880035,
|
| 2236 |
+
"logits/rejected": -0.2804562747478485,
|
| 2237 |
+
"logps/chosen": -1005.9734497070312,
|
| 2238 |
+
"logps/rejected": -1714.1412353515625,
|
| 2239 |
+
"loss": 0.0197,
|
| 2240 |
+
"rewards/accuracies": 0.971875011920929,
|
| 2241 |
+
"rewards/chosen": 8.944713592529297,
|
| 2242 |
+
"rewards/margins": 46.28833770751953,
|
| 2243 |
+
"rewards/rejected": -37.343624114990234,
|
| 2244 |
+
"step": 1490
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 9.93687707641196,
|
| 2248 |
+
"grad_norm": 0.09091484546661377,
|
| 2249 |
+
"learning_rate": 0.0,
|
| 2250 |
+
"logits/chosen": -0.21035358309745789,
|
| 2251 |
+
"logits/rejected": -0.2859784662723541,
|
| 2252 |
+
"logps/chosen": -1002.4822387695312,
|
| 2253 |
+
"logps/rejected": -1666.3056640625,
|
| 2254 |
+
"loss": 0.0272,
|
| 2255 |
+
"rewards/accuracies": 0.9609375,
|
| 2256 |
+
"rewards/chosen": 9.086991310119629,
|
| 2257 |
+
"rewards/margins": 45.62021255493164,
|
| 2258 |
+
"rewards/rejected": -36.53321838378906,
|
| 2259 |
+
"step": 1500
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 9.93687707641196,
|
| 2263 |
+
"step": 1500,
|
| 2264 |
+
"total_flos": 2.9219422466201354e+19,
|
| 2265 |
+
"train_loss": 0.07423472365736962,
|
| 2266 |
+
"train_runtime": 24807.8996,
|
| 2267 |
+
"train_samples_per_second": 3.881,
|
| 2268 |
+
"train_steps_per_second": 0.06
|
| 2269 |
+
}
|
| 2270 |
+
],
|
| 2271 |
+
"logging_steps": 10,
|
| 2272 |
+
"max_steps": 1500,
|
| 2273 |
+
"num_input_tokens_seen": 0,
|
| 2274 |
+
"num_train_epochs": 10,
|
| 2275 |
+
"save_steps": 500,
|
| 2276 |
+
"stateful_callbacks": {
|
| 2277 |
+
"TrainerControl": {
|
| 2278 |
+
"args": {
|
| 2279 |
+
"should_epoch_stop": false,
|
| 2280 |
+
"should_evaluate": false,
|
| 2281 |
+
"should_log": false,
|
| 2282 |
+
"should_save": true,
|
| 2283 |
+
"should_training_stop": true
|
| 2284 |
+
},
|
| 2285 |
+
"attributes": {}
|
| 2286 |
+
}
|
| 2287 |
+
},
|
| 2288 |
+
"total_flos": 2.9219422466201354e+19,
|
| 2289 |
+
"train_batch_size": 1,
|
| 2290 |
+
"trial_name": null,
|
| 2291 |
+
"trial_params": null
|
| 2292 |
+
}
|
training_loss.png
ADDED
|
training_rewards_accuracies.png
ADDED
|