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_from_base
|
|
@@ -17,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 17 |
|
| 18 |
# dpo_crafting_lora_from_base
|
| 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_from_base
|
|
|
|
| 18 |
|
| 19 |
# dpo_crafting_lora_from_base
|
| 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": 0.07426366006334623,
|
| 5 |
+
"train_runtime": 24708.7747,
|
| 6 |
+
"train_samples_per_second": 3.897,
|
| 7 |
+
"train_steps_per_second": 0.061
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.93687707641196,
|
| 3 |
+
"total_flos": 2.9219422466201354e+19,
|
| 4 |
+
"train_loss": 0.07426366006334623,
|
| 5 |
+
"train_runtime": 24708.7747,
|
| 6 |
+
"train_samples_per_second": 3.897,
|
| 7 |
+
"train_steps_per_second": 0.061
|
| 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": 40.747467041015625,
|
| 14 |
+
"learning_rate": 3.3333333333333335e-07,
|
| 15 |
+
"logits/chosen": -0.12511932849884033,
|
| 16 |
+
"logits/rejected": -0.18846073746681213,
|
| 17 |
+
"logps/chosen": -1117.1700439453125,
|
| 18 |
+
"logps/rejected": -1311.204345703125,
|
| 19 |
+
"loss": 0.7232,
|
| 20 |
+
"rewards/accuracies": 0.375,
|
| 21 |
+
"rewards/chosen": 0.004250399302691221,
|
| 22 |
+
"rewards/margins": -0.0051729860715568066,
|
| 23 |
+
"rewards/rejected": 0.00942337978631258,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.132890365448505,
|
| 28 |
+
"grad_norm": 63.1617546081543,
|
| 29 |
+
"learning_rate": 6.666666666666667e-07,
|
| 30 |
+
"logits/chosen": -0.1253906637430191,
|
| 31 |
+
"logits/rejected": -0.1845361292362213,
|
| 32 |
+
"logps/chosen": -1105.6373291015625,
|
| 33 |
+
"logps/rejected": -1318.8868408203125,
|
| 34 |
+
"loss": 0.7263,
|
| 35 |
+
"rewards/accuracies": 0.5031250715255737,
|
| 36 |
+
"rewards/chosen": -0.03344225138425827,
|
| 37 |
+
"rewards/margins": 0.015075692906975746,
|
| 38 |
+
"rewards/rejected": -0.04851795360445976,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.19933554817275748,
|
| 43 |
+
"grad_norm": 35.549259185791016,
|
| 44 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 45 |
+
"logits/chosen": -0.13364776968955994,
|
| 46 |
+
"logits/rejected": -0.1953742951154709,
|
| 47 |
+
"logps/chosen": -1098.7236328125,
|
| 48 |
+
"logps/rejected": -1286.4495849609375,
|
| 49 |
+
"loss": 0.7018,
|
| 50 |
+
"rewards/accuracies": 0.515625,
|
| 51 |
+
"rewards/chosen": -0.17741113901138306,
|
| 52 |
+
"rewards/margins": 0.07103082537651062,
|
| 53 |
+
"rewards/rejected": -0.24844194948673248,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.26578073089701,
|
| 58 |
+
"grad_norm": 38.198516845703125,
|
| 59 |
+
"learning_rate": 1.3333333333333334e-06,
|
| 60 |
+
"logits/chosen": -0.13651709258556366,
|
| 61 |
+
"logits/rejected": -0.19466470181941986,
|
| 62 |
+
"logps/chosen": -1057.7530517578125,
|
| 63 |
+
"logps/rejected": -1252.78173828125,
|
| 64 |
+
"loss": 0.6744,
|
| 65 |
+
"rewards/accuracies": 0.515625,
|
| 66 |
+
"rewards/chosen": -0.45364782214164734,
|
| 67 |
+
"rewards/margins": 0.16406498849391937,
|
| 68 |
+
"rewards/rejected": -0.6177127957344055,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.33222591362126247,
|
| 73 |
+
"grad_norm": 27.63871955871582,
|
| 74 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 75 |
+
"logits/chosen": -0.13217487931251526,
|
| 76 |
+
"logits/rejected": -0.19238920509815216,
|
| 77 |
+
"logps/chosen": -1069.2860107421875,
|
| 78 |
+
"logps/rejected": -1261.974853515625,
|
| 79 |
+
"loss": 0.6492,
|
| 80 |
+
"rewards/accuracies": 0.589062511920929,
|
| 81 |
+
"rewards/chosen": -0.8812439441680908,
|
| 82 |
+
"rewards/margins": 0.36069372296333313,
|
| 83 |
+
"rewards/rejected": -1.2419377565383911,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.39867109634551495,
|
| 88 |
+
"grad_norm": 21.243654251098633,
|
| 89 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 90 |
+
"logits/chosen": -0.13583233952522278,
|
| 91 |
+
"logits/rejected": -0.1979021281003952,
|
| 92 |
+
"logps/chosen": -1059.2923583984375,
|
| 93 |
+
"logps/rejected": -1271.649169921875,
|
| 94 |
+
"loss": 0.6293,
|
| 95 |
+
"rewards/accuracies": 0.6343749761581421,
|
| 96 |
+
"rewards/chosen": -0.9827756881713867,
|
| 97 |
+
"rewards/margins": 0.5085424780845642,
|
| 98 |
+
"rewards/rejected": -1.4913183450698853,
|
| 99 |
+
"step": 60
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.46511627906976744,
|
| 103 |
+
"grad_norm": 19.875577926635742,
|
| 104 |
+
"learning_rate": 2.3333333333333336e-06,
|
| 105 |
+
"logits/chosen": -0.13559813797473907,
|
| 106 |
+
"logits/rejected": -0.19648894667625427,
|
| 107 |
+
"logps/chosen": -1103.181396484375,
|
| 108 |
+
"logps/rejected": -1314.6522216796875,
|
| 109 |
+
"loss": 0.5424,
|
| 110 |
+
"rewards/accuracies": 0.707812488079071,
|
| 111 |
+
"rewards/chosen": -0.8136237859725952,
|
| 112 |
+
"rewards/margins": 0.6972610950469971,
|
| 113 |
+
"rewards/rejected": -1.5108850002288818,
|
| 114 |
+
"step": 70
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.53156146179402,
|
| 118 |
+
"grad_norm": 20.229171752929688,
|
| 119 |
+
"learning_rate": 2.666666666666667e-06,
|
| 120 |
+
"logits/chosen": -0.11205008625984192,
|
| 121 |
+
"logits/rejected": -0.19006861746311188,
|
| 122 |
+
"logps/chosen": -1065.726806640625,
|
| 123 |
+
"logps/rejected": -1319.034423828125,
|
| 124 |
+
"loss": 0.4856,
|
| 125 |
+
"rewards/accuracies": 0.7828124761581421,
|
| 126 |
+
"rewards/chosen": -0.6954863667488098,
|
| 127 |
+
"rewards/margins": 1.0426087379455566,
|
| 128 |
+
"rewards/rejected": -1.7380950450897217,
|
| 129 |
+
"step": 80
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.5980066445182725,
|
| 133 |
+
"grad_norm": 17.880022048950195,
|
| 134 |
+
"learning_rate": 3e-06,
|
| 135 |
+
"logits/chosen": -0.12407304346561432,
|
| 136 |
+
"logits/rejected": -0.18850643932819366,
|
| 137 |
+
"logps/chosen": -1076.2564697265625,
|
| 138 |
+
"logps/rejected": -1331.6634521484375,
|
| 139 |
+
"loss": 0.4188,
|
| 140 |
+
"rewards/accuracies": 0.840624988079071,
|
| 141 |
+
"rewards/chosen": -0.3792995810508728,
|
| 142 |
+
"rewards/margins": 1.265782117843628,
|
| 143 |
+
"rewards/rejected": -1.645081639289856,
|
| 144 |
+
"step": 90
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.6644518272425249,
|
| 148 |
+
"grad_norm": 12.63503646850586,
|
| 149 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 150 |
+
"logits/chosen": -0.121916763484478,
|
| 151 |
+
"logits/rejected": -0.19661013782024384,
|
| 152 |
+
"logps/chosen": -1058.5997314453125,
|
| 153 |
+
"logps/rejected": -1334.9495849609375,
|
| 154 |
+
"loss": 0.3597,
|
| 155 |
+
"rewards/accuracies": 0.8609374761581421,
|
| 156 |
+
"rewards/chosen": -0.5200275182723999,
|
| 157 |
+
"rewards/margins": 1.9380321502685547,
|
| 158 |
+
"rewards/rejected": -2.458059549331665,
|
| 159 |
+
"step": 100
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.7308970099667774,
|
| 163 |
+
"grad_norm": 14.828380584716797,
|
| 164 |
+
"learning_rate": 3.6666666666666666e-06,
|
| 165 |
+
"logits/chosen": -0.13187702000141144,
|
| 166 |
+
"logits/rejected": -0.19965648651123047,
|
| 167 |
+
"logps/chosen": -1074.6199951171875,
|
| 168 |
+
"logps/rejected": -1312.4927978515625,
|
| 169 |
+
"loss": 0.2984,
|
| 170 |
+
"rewards/accuracies": 0.90625,
|
| 171 |
+
"rewards/chosen": -0.4199390709400177,
|
| 172 |
+
"rewards/margins": 2.6119236946105957,
|
| 173 |
+
"rewards/rejected": -3.031863212585449,
|
| 174 |
+
"step": 110
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.7973421926910299,
|
| 178 |
+
"grad_norm": 10.107036590576172,
|
| 179 |
+
"learning_rate": 4.000000000000001e-06,
|
| 180 |
+
"logits/chosen": -0.12684178352355957,
|
| 181 |
+
"logits/rejected": -0.1979806274175644,
|
| 182 |
+
"logps/chosen": -1091.7188720703125,
|
| 183 |
+
"logps/rejected": -1349.757568359375,
|
| 184 |
+
"loss": 0.2236,
|
| 185 |
+
"rewards/accuracies": 0.9312500357627869,
|
| 186 |
+
"rewards/chosen": -0.07451216131448746,
|
| 187 |
+
"rewards/margins": 4.150519847869873,
|
| 188 |
+
"rewards/rejected": -4.225031852722168,
|
| 189 |
+
"step": 120
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.8637873754152824,
|
| 193 |
+
"grad_norm": 4.532182693481445,
|
| 194 |
+
"learning_rate": 4.333333333333334e-06,
|
| 195 |
+
"logits/chosen": -0.13108162581920624,
|
| 196 |
+
"logits/rejected": -0.19549141824245453,
|
| 197 |
+
"logps/chosen": -1100.24951171875,
|
| 198 |
+
"logps/rejected": -1363.146728515625,
|
| 199 |
+
"loss": 0.1608,
|
| 200 |
+
"rewards/accuracies": 0.9468750357627869,
|
| 201 |
+
"rewards/chosen": 0.3449128270149231,
|
| 202 |
+
"rewards/margins": 5.669872760772705,
|
| 203 |
+
"rewards/rejected": -5.324959754943848,
|
| 204 |
+
"step": 130
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.9302325581395349,
|
| 208 |
+
"grad_norm": 2.4136996269226074,
|
| 209 |
+
"learning_rate": 4.666666666666667e-06,
|
| 210 |
+
"logits/chosen": -0.13807465136051178,
|
| 211 |
+
"logits/rejected": -0.20656287670135498,
|
| 212 |
+
"logps/chosen": -1032.328369140625,
|
| 213 |
+
"logps/rejected": -1357.612548828125,
|
| 214 |
+
"loss": 0.1282,
|
| 215 |
+
"rewards/accuracies": 0.9437499642372131,
|
| 216 |
+
"rewards/chosen": 1.2508221864700317,
|
| 217 |
+
"rewards/margins": 7.064514636993408,
|
| 218 |
+
"rewards/rejected": -5.813692092895508,
|
| 219 |
+
"step": 140
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.9966777408637874,
|
| 223 |
+
"grad_norm": 22.84921646118164,
|
| 224 |
+
"learning_rate": 5e-06,
|
| 225 |
+
"logits/chosen": -0.1460917443037033,
|
| 226 |
+
"logits/rejected": -0.21644163131713867,
|
| 227 |
+
"logps/chosen": -1055.5206298828125,
|
| 228 |
+
"logps/rejected": -1391.220947265625,
|
| 229 |
+
"loss": 0.1053,
|
| 230 |
+
"rewards/accuracies": 0.9375,
|
| 231 |
+
"rewards/chosen": 0.7099586725234985,
|
| 232 |
+
"rewards/margins": 9.3050537109375,
|
| 233 |
+
"rewards/rejected": -8.59509563446045,
|
| 234 |
+
"step": 150
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 1.0598006644518272,
|
| 238 |
+
"grad_norm": 5.813467502593994,
|
| 239 |
+
"learning_rate": 4.999323102948655e-06,
|
| 240 |
+
"logits/chosen": -0.14507660269737244,
|
| 241 |
+
"logits/rejected": -0.21615180373191833,
|
| 242 |
+
"logps/chosen": -1054.1131591796875,
|
| 243 |
+
"logps/rejected": -1392.970947265625,
|
| 244 |
+
"loss": 0.0871,
|
| 245 |
+
"rewards/accuracies": 0.9407894611358643,
|
| 246 |
+
"rewards/chosen": 1.4173532724380493,
|
| 247 |
+
"rewards/margins": 11.686441421508789,
|
| 248 |
+
"rewards/rejected": -10.269088745117188,
|
| 249 |
+
"step": 160
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.1262458471760797,
|
| 253 |
+
"grad_norm": 25.732624053955078,
|
| 254 |
+
"learning_rate": 4.997292778346312e-06,
|
| 255 |
+
"logits/chosen": -0.1539006382226944,
|
| 256 |
+
"logits/rejected": -0.21280081570148468,
|
| 257 |
+
"logps/chosen": -1064.2578125,
|
| 258 |
+
"logps/rejected": -1432.212646484375,
|
| 259 |
+
"loss": 0.0596,
|
| 260 |
+
"rewards/accuracies": 0.9609375,
|
| 261 |
+
"rewards/chosen": 1.7107740640640259,
|
| 262 |
+
"rewards/margins": 15.027220726013184,
|
| 263 |
+
"rewards/rejected": -13.316445350646973,
|
| 264 |
+
"step": 170
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 1.1926910299003322,
|
| 268 |
+
"grad_norm": 2.911512851715088,
|
| 269 |
+
"learning_rate": 4.993910125649561e-06,
|
| 270 |
+
"logits/chosen": -0.13852202892303467,
|
| 271 |
+
"logits/rejected": -0.2171792984008789,
|
| 272 |
+
"logps/chosen": -992.4456787109375,
|
| 273 |
+
"logps/rejected": -1381.4122314453125,
|
| 274 |
+
"loss": 0.0456,
|
| 275 |
+
"rewards/accuracies": 0.964062511920929,
|
| 276 |
+
"rewards/chosen": 3.4778242111206055,
|
| 277 |
+
"rewards/margins": 15.223333358764648,
|
| 278 |
+
"rewards/rejected": -11.745508193969727,
|
| 279 |
+
"step": 180
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 1.2591362126245846,
|
| 283 |
+
"grad_norm": 0.5809551477432251,
|
| 284 |
+
"learning_rate": 4.989176976624511e-06,
|
| 285 |
+
"logits/chosen": -0.16654856503009796,
|
| 286 |
+
"logits/rejected": -0.23170927166938782,
|
| 287 |
+
"logps/chosen": -1053.9530029296875,
|
| 288 |
+
"logps/rejected": -1434.646484375,
|
| 289 |
+
"loss": 0.0562,
|
| 290 |
+
"rewards/accuracies": 0.940625011920929,
|
| 291 |
+
"rewards/chosen": 3.07829213142395,
|
| 292 |
+
"rewards/margins": 18.499500274658203,
|
| 293 |
+
"rewards/rejected": -15.421209335327148,
|
| 294 |
+
"step": 190
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 1.3255813953488373,
|
| 298 |
+
"grad_norm": 4.461428165435791,
|
| 299 |
+
"learning_rate": 4.983095894354858e-06,
|
| 300 |
+
"logits/chosen": -0.17254643142223358,
|
| 301 |
+
"logits/rejected": -0.23220513761043549,
|
| 302 |
+
"logps/chosen": -1038.1956787109375,
|
| 303 |
+
"logps/rejected": -1425.747802734375,
|
| 304 |
+
"loss": 0.0551,
|
| 305 |
+
"rewards/accuracies": 0.9453125,
|
| 306 |
+
"rewards/chosen": 4.759212493896484,
|
| 307 |
+
"rewards/margins": 18.334705352783203,
|
| 308 |
+
"rewards/rejected": -13.575494766235352,
|
| 309 |
+
"step": 200
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 1.3920265780730898,
|
| 313 |
+
"grad_norm": 0.13902103900909424,
|
| 314 |
+
"learning_rate": 4.975670171853926e-06,
|
| 315 |
+
"logits/chosen": -0.1616649478673935,
|
| 316 |
+
"logits/rejected": -0.22982370853424072,
|
| 317 |
+
"logps/chosen": -1068.1690673828125,
|
| 318 |
+
"logps/rejected": -1542.0587158203125,
|
| 319 |
+
"loss": 0.0766,
|
| 320 |
+
"rewards/accuracies": 0.9515624642372131,
|
| 321 |
+
"rewards/chosen": 1.5794833898544312,
|
| 322 |
+
"rewards/margins": 24.19274139404297,
|
| 323 |
+
"rewards/rejected": -22.613258361816406,
|
| 324 |
+
"step": 210
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.4584717607973423,
|
| 328 |
+
"grad_norm": 111.07405090332031,
|
| 329 |
+
"learning_rate": 4.966903830281449e-06,
|
| 330 |
+
"logits/chosen": -0.1702309548854828,
|
| 331 |
+
"logits/rejected": -0.23228994011878967,
|
| 332 |
+
"logps/chosen": -1009.6397705078125,
|
| 333 |
+
"logps/rejected": -1450.9083251953125,
|
| 334 |
+
"loss": 0.0614,
|
| 335 |
+
"rewards/accuracies": 0.9468750357627869,
|
| 336 |
+
"rewards/chosen": 4.363029479980469,
|
| 337 |
+
"rewards/margins": 21.61128044128418,
|
| 338 |
+
"rewards/rejected": -17.248252868652344,
|
| 339 |
+
"step": 220
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.5249169435215948,
|
| 343 |
+
"grad_norm": 1.088158369064331,
|
| 344 |
+
"learning_rate": 4.956801616766033e-06,
|
| 345 |
+
"logits/chosen": -0.17264895141124725,
|
| 346 |
+
"logits/rejected": -0.22918781638145447,
|
| 347 |
+
"logps/chosen": -1067.2901611328125,
|
| 348 |
+
"logps/rejected": -1552.646728515625,
|
| 349 |
+
"loss": 0.0404,
|
| 350 |
+
"rewards/accuracies": 0.956250011920929,
|
| 351 |
+
"rewards/chosen": 3.915598154067993,
|
| 352 |
+
"rewards/margins": 26.750593185424805,
|
| 353 |
+
"rewards/rejected": -22.83499526977539,
|
| 354 |
+
"step": 230
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.5913621262458473,
|
| 358 |
+
"grad_norm": 1.4660481214523315,
|
| 359 |
+
"learning_rate": 4.9453690018345144e-06,
|
| 360 |
+
"logits/chosen": -0.17638970911502838,
|
| 361 |
+
"logits/rejected": -0.23336604237556458,
|
| 362 |
+
"logps/chosen": -1033.65771484375,
|
| 363 |
+
"logps/rejected": -1489.400146484375,
|
| 364 |
+
"loss": 0.0358,
|
| 365 |
+
"rewards/accuracies": 0.957812488079071,
|
| 366 |
+
"rewards/chosen": 4.591525554656982,
|
| 367 |
+
"rewards/margins": 24.92582893371582,
|
| 368 |
+
"rewards/rejected": -20.334304809570312,
|
| 369 |
+
"step": 240
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.6578073089700998,
|
| 373 |
+
"grad_norm": 0.6518616080284119,
|
| 374 |
+
"learning_rate": 4.93261217644956e-06,
|
| 375 |
+
"logits/chosen": -0.16551312804222107,
|
| 376 |
+
"logits/rejected": -0.2351105809211731,
|
| 377 |
+
"logps/chosen": -1069.6429443359375,
|
| 378 |
+
"logps/rejected": -1581.255126953125,
|
| 379 |
+
"loss": 0.0554,
|
| 380 |
+
"rewards/accuracies": 0.9656250476837158,
|
| 381 |
+
"rewards/chosen": 2.4628138542175293,
|
| 382 |
+
"rewards/margins": 27.925844192504883,
|
| 383 |
+
"rewards/rejected": -25.46302604675293,
|
| 384 |
+
"step": 250
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.7242524916943522,
|
| 388 |
+
"grad_norm": 0.4813426434993744,
|
| 389 |
+
"learning_rate": 4.91853804865716e-06,
|
| 390 |
+
"logits/chosen": -0.18384608626365662,
|
| 391 |
+
"logits/rejected": -0.24281182885169983,
|
| 392 |
+
"logps/chosen": -968.432861328125,
|
| 393 |
+
"logps/rejected": -1347.4447021484375,
|
| 394 |
+
"loss": 0.0351,
|
| 395 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 396 |
+
"rewards/chosen": 7.137730121612549,
|
| 397 |
+
"rewards/margins": 21.095685958862305,
|
| 398 |
+
"rewards/rejected": -13.957955360412598,
|
| 399 |
+
"step": 260
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.7906976744186047,
|
| 403 |
+
"grad_norm": 11.40811824798584,
|
| 404 |
+
"learning_rate": 4.903154239845798e-06,
|
| 405 |
+
"logits/chosen": -0.1812049150466919,
|
| 406 |
+
"logits/rejected": -0.25064709782600403,
|
| 407 |
+
"logps/chosen": -1056.5556640625,
|
| 408 |
+
"logps/rejected": -1586.718017578125,
|
| 409 |
+
"loss": 0.0393,
|
| 410 |
+
"rewards/accuracies": 0.953125,
|
| 411 |
+
"rewards/chosen": 3.66015625,
|
| 412 |
+
"rewards/margins": 30.29279899597168,
|
| 413 |
+
"rewards/rejected": -26.632644653320312,
|
| 414 |
+
"step": 270
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 1.8571428571428572,
|
| 418 |
+
"grad_norm": 0.7297212481498718,
|
| 419 |
+
"learning_rate": 4.88646908061933e-06,
|
| 420 |
+
"logits/chosen": -0.18657241761684418,
|
| 421 |
+
"logits/rejected": -0.2551916539669037,
|
| 422 |
+
"logps/chosen": -968.1968383789062,
|
| 423 |
+
"logps/rejected": -1500.6644287109375,
|
| 424 |
+
"loss": 0.0391,
|
| 425 |
+
"rewards/accuracies": 0.948437511920929,
|
| 426 |
+
"rewards/chosen": 6.839397430419922,
|
| 427 |
+
"rewards/margins": 27.143301010131836,
|
| 428 |
+
"rewards/rejected": -20.303905487060547,
|
| 429 |
+
"step": 280
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.9235880398671097,
|
| 433 |
+
"grad_norm": 0.17506621778011322,
|
| 434 |
+
"learning_rate": 4.868491606285823e-06,
|
| 435 |
+
"logits/chosen": -0.19025684893131256,
|
| 436 |
+
"logits/rejected": -0.25329911708831787,
|
| 437 |
+
"logps/chosen": -1026.042236328125,
|
| 438 |
+
"logps/rejected": -1535.4869384765625,
|
| 439 |
+
"loss": 0.0436,
|
| 440 |
+
"rewards/accuracies": 0.9531250596046448,
|
| 441 |
+
"rewards/chosen": 6.215095043182373,
|
| 442 |
+
"rewards/margins": 29.239473342895508,
|
| 443 |
+
"rewards/rejected": -23.024375915527344,
|
| 444 |
+
"step": 290
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.9900332225913622,
|
| 448 |
+
"grad_norm": 0.1333008110523224,
|
| 449 |
+
"learning_rate": 4.849231551964771e-06,
|
| 450 |
+
"logits/chosen": -0.1948496401309967,
|
| 451 |
+
"logits/rejected": -0.25613006949424744,
|
| 452 |
+
"logps/chosen": -1025.02783203125,
|
| 453 |
+
"logps/rejected": -1517.8956298828125,
|
| 454 |
+
"loss": 0.0381,
|
| 455 |
+
"rewards/accuracies": 0.9546875953674316,
|
| 456 |
+
"rewards/chosen": 7.619020938873291,
|
| 457 |
+
"rewards/margins": 27.76497459411621,
|
| 458 |
+
"rewards/rejected": -20.145952224731445,
|
| 459 |
+
"step": 300
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 2.053156146179402,
|
| 463 |
+
"grad_norm": 1.3418188095092773,
|
| 464 |
+
"learning_rate": 4.828699347315357e-06,
|
| 465 |
+
"logits/chosen": -0.17470988631248474,
|
| 466 |
+
"logits/rejected": -0.2475568950176239,
|
| 467 |
+
"logps/chosen": -994.5497436523438,
|
| 468 |
+
"logps/rejected": -1511.7833251953125,
|
| 469 |
+
"loss": 0.0276,
|
| 470 |
+
"rewards/accuracies": 0.9621711373329163,
|
| 471 |
+
"rewards/chosen": 7.3509602546691895,
|
| 472 |
+
"rewards/margins": 28.877328872680664,
|
| 473 |
+
"rewards/rejected": -21.526369094848633,
|
| 474 |
+
"step": 310
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 2.1196013289036544,
|
| 478 |
+
"grad_norm": 0.7398647665977478,
|
| 479 |
+
"learning_rate": 4.806906110888606e-06,
|
| 480 |
+
"logits/chosen": -0.19838787615299225,
|
| 481 |
+
"logits/rejected": -0.253930926322937,
|
| 482 |
+
"logps/chosen": -1008.02734375,
|
| 483 |
+
"logps/rejected": -1496.8922119140625,
|
| 484 |
+
"loss": 0.0344,
|
| 485 |
+
"rewards/accuracies": 0.953125,
|
| 486 |
+
"rewards/chosen": 6.126870155334473,
|
| 487 |
+
"rewards/margins": 30.307435989379883,
|
| 488 |
+
"rewards/rejected": -24.180564880371094,
|
| 489 |
+
"step": 320
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 2.186046511627907,
|
| 493 |
+
"grad_norm": 0.2915615439414978,
|
| 494 |
+
"learning_rate": 4.783863644106502e-06,
|
| 495 |
+
"logits/chosen": -0.18888217210769653,
|
| 496 |
+
"logits/rejected": -0.2543439567089081,
|
| 497 |
+
"logps/chosen": -1001.6741333007812,
|
| 498 |
+
"logps/rejected": -1547.251953125,
|
| 499 |
+
"loss": 0.0294,
|
| 500 |
+
"rewards/accuracies": 0.9593749642372131,
|
| 501 |
+
"rewards/chosen": 6.256338119506836,
|
| 502 |
+
"rewards/margins": 32.92168426513672,
|
| 503 |
+
"rewards/rejected": -26.66534423828125,
|
| 504 |
+
"step": 330
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 2.2524916943521593,
|
| 508 |
+
"grad_norm": 0.44617852568626404,
|
| 509 |
+
"learning_rate": 4.759584424871302e-06,
|
| 510 |
+
"logits/chosen": -0.19481198489665985,
|
| 511 |
+
"logits/rejected": -0.2515893876552582,
|
| 512 |
+
"logps/chosen": -1063.011474609375,
|
| 513 |
+
"logps/rejected": -1607.3277587890625,
|
| 514 |
+
"loss": 0.0264,
|
| 515 |
+
"rewards/accuracies": 0.964062511920929,
|
| 516 |
+
"rewards/chosen": 7.461002826690674,
|
| 517 |
+
"rewards/margins": 33.48613357543945,
|
| 518 |
+
"rewards/rejected": -26.025131225585938,
|
| 519 |
+
"step": 340
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 2.318936877076412,
|
| 523 |
+
"grad_norm": 0.5236634016036987,
|
| 524 |
+
"learning_rate": 4.734081600808531e-06,
|
| 525 |
+
"logits/chosen": -0.2034038007259369,
|
| 526 |
+
"logits/rejected": -0.2618187367916107,
|
| 527 |
+
"logps/chosen": -1072.1470947265625,
|
| 528 |
+
"logps/rejected": -1632.1077880859375,
|
| 529 |
+
"loss": 0.0389,
|
| 530 |
+
"rewards/accuracies": 0.9468750357627869,
|
| 531 |
+
"rewards/chosen": 6.423084259033203,
|
| 532 |
+
"rewards/margins": 35.269798278808594,
|
| 533 |
+
"rewards/rejected": -28.846710205078125,
|
| 534 |
+
"step": 350
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 2.3853820598006643,
|
| 538 |
+
"grad_norm": 0.05544855445623398,
|
| 539 |
+
"learning_rate": 4.707368982147318e-06,
|
| 540 |
+
"logits/chosen": -0.18389806151390076,
|
| 541 |
+
"logits/rejected": -0.2554208040237427,
|
| 542 |
+
"logps/chosen": -976.6298828125,
|
| 543 |
+
"logps/rejected": -1534.483154296875,
|
| 544 |
+
"loss": 0.0313,
|
| 545 |
+
"rewards/accuracies": 0.9578125476837158,
|
| 546 |
+
"rewards/chosen": 6.575656414031982,
|
| 547 |
+
"rewards/margins": 32.54267120361328,
|
| 548 |
+
"rewards/rejected": -25.967016220092773,
|
| 549 |
+
"step": 360
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 2.451827242524917,
|
| 553 |
+
"grad_norm": 0.2025989443063736,
|
| 554 |
+
"learning_rate": 4.679461034241906e-06,
|
| 555 |
+
"logits/chosen": -0.1815650463104248,
|
| 556 |
+
"logits/rejected": -0.2540404796600342,
|
| 557 |
+
"logps/chosen": -1047.0966796875,
|
| 558 |
+
"logps/rejected": -1622.110595703125,
|
| 559 |
+
"loss": 0.0276,
|
| 560 |
+
"rewards/accuracies": 0.9625000953674316,
|
| 561 |
+
"rewards/chosen": 6.719258785247803,
|
| 562 |
+
"rewards/margins": 35.81674575805664,
|
| 563 |
+
"rewards/rejected": -29.097484588623047,
|
| 564 |
+
"step": 370
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 2.5182724252491693,
|
| 568 |
+
"grad_norm": 0.7222418189048767,
|
| 569 |
+
"learning_rate": 4.650372869738415e-06,
|
| 570 |
+
"logits/chosen": -0.18268240988254547,
|
| 571 |
+
"logits/rejected": -0.2617727220058441,
|
| 572 |
+
"logps/chosen": -965.8853759765625,
|
| 573 |
+
"logps/rejected": -1582.7860107421875,
|
| 574 |
+
"loss": 0.0332,
|
| 575 |
+
"rewards/accuracies": 0.956250011920929,
|
| 576 |
+
"rewards/chosen": 7.1367573738098145,
|
| 577 |
+
"rewards/margins": 33.75447082519531,
|
| 578 |
+
"rewards/rejected": -26.617713928222656,
|
| 579 |
+
"step": 380
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 2.584717607973422,
|
| 583 |
+
"grad_norm": 0.1559152901172638,
|
| 584 |
+
"learning_rate": 4.620120240391065e-06,
|
| 585 |
+
"logits/chosen": -0.19453352689743042,
|
| 586 |
+
"logits/rejected": -0.2535034418106079,
|
| 587 |
+
"logps/chosen": -1019.660888671875,
|
| 588 |
+
"logps/rejected": -1637.6041259765625,
|
| 589 |
+
"loss": 0.032,
|
| 590 |
+
"rewards/accuracies": 0.9640625715255737,
|
| 591 |
+
"rewards/chosen": 6.416092395782471,
|
| 592 |
+
"rewards/margins": 36.01222610473633,
|
| 593 |
+
"rewards/rejected": -29.596132278442383,
|
| 594 |
+
"step": 390
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 2.6511627906976747,
|
| 598 |
+
"grad_norm": 0.05074378848075867,
|
| 599 |
+
"learning_rate": 4.588719528532342e-06,
|
| 600 |
+
"logits/chosen": -0.1765459030866623,
|
| 601 |
+
"logits/rejected": -0.259499192237854,
|
| 602 |
+
"logps/chosen": -942.0123291015625,
|
| 603 |
+
"logps/rejected": -1461.3004150390625,
|
| 604 |
+
"loss": 0.0334,
|
| 605 |
+
"rewards/accuracies": 0.956250011920929,
|
| 606 |
+
"rewards/chosen": 8.615708351135254,
|
| 607 |
+
"rewards/margins": 29.26686668395996,
|
| 608 |
+
"rewards/rejected": -20.651159286499023,
|
| 609 |
+
"step": 400
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 2.717607973421927,
|
| 613 |
+
"grad_norm": 0.027149733155965805,
|
| 614 |
+
"learning_rate": 4.556187738201656e-06,
|
| 615 |
+
"logits/chosen": -0.19329792261123657,
|
| 616 |
+
"logits/rejected": -0.2614816427230835,
|
| 617 |
+
"logps/chosen": -974.2481689453125,
|
| 618 |
+
"logps/rejected": -1543.07177734375,
|
| 619 |
+
"loss": 0.0366,
|
| 620 |
+
"rewards/accuracies": 0.949999988079071,
|
| 621 |
+
"rewards/chosen": 5.765076637268066,
|
| 622 |
+
"rewards/margins": 35.13932418823242,
|
| 623 |
+
"rewards/rejected": -29.37424087524414,
|
| 624 |
+
"step": 410
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 2.7840531561461797,
|
| 628 |
+
"grad_norm": 0.07503657788038254,
|
| 629 |
+
"learning_rate": 4.522542485937369e-06,
|
| 630 |
+
"logits/chosen": -0.19822394847869873,
|
| 631 |
+
"logits/rejected": -0.2589871287345886,
|
| 632 |
+
"logps/chosen": -1010.1796264648438,
|
| 633 |
+
"logps/rejected": -1573.482421875,
|
| 634 |
+
"loss": 0.0347,
|
| 635 |
+
"rewards/accuracies": 0.9562500715255737,
|
| 636 |
+
"rewards/chosen": 6.699889183044434,
|
| 637 |
+
"rewards/margins": 35.83893966674805,
|
| 638 |
+
"rewards/rejected": -29.13905143737793,
|
| 639 |
+
"step": 420
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 2.850498338870432,
|
| 643 |
+
"grad_norm": 0.08216100186109543,
|
| 644 |
+
"learning_rate": 4.48780199123712e-06,
|
| 645 |
+
"logits/chosen": -0.19484762847423553,
|
| 646 |
+
"logits/rejected": -0.2583960294723511,
|
| 647 |
+
"logps/chosen": -1018.3121337890625,
|
| 648 |
+
"logps/rejected": -1550.55615234375,
|
| 649 |
+
"loss": 0.0327,
|
| 650 |
+
"rewards/accuracies": 0.9546875357627869,
|
| 651 |
+
"rewards/chosen": 8.480841636657715,
|
| 652 |
+
"rewards/margins": 33.863243103027344,
|
| 653 |
+
"rewards/rejected": -25.382402420043945,
|
| 654 |
+
"step": 430
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 2.9169435215946846,
|
| 658 |
+
"grad_norm": 0.04343642294406891,
|
| 659 |
+
"learning_rate": 4.451985066691649e-06,
|
| 660 |
+
"logits/chosen": -0.20001523196697235,
|
| 661 |
+
"logits/rejected": -0.26353001594543457,
|
| 662 |
+
"logps/chosen": -979.586669921875,
|
| 663 |
+
"logps/rejected": -1509.11328125,
|
| 664 |
+
"loss": 0.0313,
|
| 665 |
+
"rewards/accuracies": 0.956250011920929,
|
| 666 |
+
"rewards/chosen": 8.662511825561523,
|
| 667 |
+
"rewards/margins": 32.48591995239258,
|
| 668 |
+
"rewards/rejected": -23.823410034179688,
|
| 669 |
+
"step": 440
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.983388704318937,
|
| 673 |
+
"grad_norm": 0.056994736194610596,
|
| 674 |
+
"learning_rate": 4.415111107797445e-06,
|
| 675 |
+
"logits/chosen": -0.1883145421743393,
|
| 676 |
+
"logits/rejected": -0.2559402883052826,
|
| 677 |
+
"logps/chosen": -1012.8984375,
|
| 678 |
+
"logps/rejected": -1519.9129638671875,
|
| 679 |
+
"loss": 0.0264,
|
| 680 |
+
"rewards/accuracies": 0.9656250476837158,
|
| 681 |
+
"rewards/chosen": 8.099644660949707,
|
| 682 |
+
"rewards/margins": 34.30991744995117,
|
| 683 |
+
"rewards/rejected": -26.21027183532715,
|
| 684 |
+
"step": 450
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 3.046511627906977,
|
| 688 |
+
"grad_norm": 0.027808276936411858,
|
| 689 |
+
"learning_rate": 4.377200082453748e-06,
|
| 690 |
+
"logits/chosen": -0.18382969498634338,
|
| 691 |
+
"logits/rejected": -0.266806423664093,
|
| 692 |
+
"logps/chosen": -933.7220458984375,
|
| 693 |
+
"logps/rejected": -1490.690673828125,
|
| 694 |
+
"loss": 0.0311,
|
| 695 |
+
"rewards/accuracies": 0.9539474248886108,
|
| 696 |
+
"rewards/chosen": 7.497554779052734,
|
| 697 |
+
"rewards/margins": 33.65736770629883,
|
| 698 |
+
"rewards/rejected": -26.159814834594727,
|
| 699 |
+
"step": 460
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 3.1129568106312293,
|
| 703 |
+
"grad_norm": 0.026268087327480316,
|
| 704 |
+
"learning_rate": 4.338272520149572e-06,
|
| 705 |
+
"logits/chosen": -0.20292654633522034,
|
| 706 |
+
"logits/rejected": -0.27201807498931885,
|
| 707 |
+
"logps/chosen": -1006.4375,
|
| 708 |
+
"logps/rejected": -1589.052734375,
|
| 709 |
+
"loss": 0.033,
|
| 710 |
+
"rewards/accuracies": 0.953125,
|
| 711 |
+
"rewards/chosen": 8.268844604492188,
|
| 712 |
+
"rewards/margins": 35.16604232788086,
|
| 713 |
+
"rewards/rejected": -26.89719581604004,
|
| 714 |
+
"step": 470
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 3.179401993355482,
|
| 718 |
+
"grad_norm": 0.014455691911280155,
|
| 719 |
+
"learning_rate": 4.2983495008466285e-06,
|
| 720 |
+
"logits/chosen": -0.17968863248825073,
|
| 721 |
+
"logits/rejected": -0.26512834429740906,
|
| 722 |
+
"logps/chosen": -972.7686157226562,
|
| 723 |
+
"logps/rejected": -1628.4017333984375,
|
| 724 |
+
"loss": 0.027,
|
| 725 |
+
"rewards/accuracies": 0.9625000357627869,
|
| 726 |
+
"rewards/chosen": 7.220202445983887,
|
| 727 |
+
"rewards/margins": 38.235836029052734,
|
| 728 |
+
"rewards/rejected": -31.01563262939453,
|
| 729 |
+
"step": 480
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 3.2458471760797343,
|
| 733 |
+
"grad_norm": 1.3661887645721436,
|
| 734 |
+
"learning_rate": 4.257452643564155e-06,
|
| 735 |
+
"logits/chosen": -0.1976873278617859,
|
| 736 |
+
"logits/rejected": -0.26860570907592773,
|
| 737 |
+
"logps/chosen": -957.9434814453125,
|
| 738 |
+
"logps/rejected": -1552.618408203125,
|
| 739 |
+
"loss": 0.0349,
|
| 740 |
+
"rewards/accuracies": 0.953125,
|
| 741 |
+
"rewards/chosen": 7.590810775756836,
|
| 742 |
+
"rewards/margins": 35.01844787597656,
|
| 743 |
+
"rewards/rejected": -27.427637100219727,
|
| 744 |
+
"step": 490
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 3.3122923588039868,
|
| 748 |
+
"grad_norm": 0.034927744418382645,
|
| 749 |
+
"learning_rate": 4.215604094671835e-06,
|
| 750 |
+
"logits/chosen": -0.2104714959859848,
|
| 751 |
+
"logits/rejected": -0.26536011695861816,
|
| 752 |
+
"logps/chosen": -1035.5135498046875,
|
| 753 |
+
"logps/rejected": -1573.6461181640625,
|
| 754 |
+
"loss": 0.0265,
|
| 755 |
+
"rewards/accuracies": 0.9625000357627869,
|
| 756 |
+
"rewards/chosen": 9.14853286743164,
|
| 757 |
+
"rewards/margins": 35.578651428222656,
|
| 758 |
+
"rewards/rejected": -26.43012046813965,
|
| 759 |
+
"step": 500
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 3.3787375415282392,
|
| 763 |
+
"grad_norm": 0.05187850445508957,
|
| 764 |
+
"learning_rate": 4.172826515897146e-06,
|
| 765 |
+
"logits/chosen": -0.19994229078292847,
|
| 766 |
+
"logits/rejected": -0.26754456758499146,
|
| 767 |
+
"logps/chosen": -997.74853515625,
|
| 768 |
+
"logps/rejected": -1574.661865234375,
|
| 769 |
+
"loss": 0.0233,
|
| 770 |
+
"rewards/accuracies": 0.9671874642372131,
|
| 771 |
+
"rewards/chosen": 8.397761344909668,
|
| 772 |
+
"rewards/margins": 34.895263671875,
|
| 773 |
+
"rewards/rejected": -26.49749755859375,
|
| 774 |
+
"step": 510
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 3.4451827242524917,
|
| 778 |
+
"grad_norm": 0.2731926739215851,
|
| 779 |
+
"learning_rate": 4.129143072053639e-06,
|
| 780 |
+
"logits/chosen": -0.19184930622577667,
|
| 781 |
+
"logits/rejected": -0.26313984394073486,
|
| 782 |
+
"logps/chosen": -1033.8189697265625,
|
| 783 |
+
"logps/rejected": -1616.736572265625,
|
| 784 |
+
"loss": 0.0222,
|
| 785 |
+
"rewards/accuracies": 0.96875,
|
| 786 |
+
"rewards/chosen": 8.199491500854492,
|
| 787 |
+
"rewards/margins": 37.13923645019531,
|
| 788 |
+
"rewards/rejected": -28.939746856689453,
|
| 789 |
+
"step": 520
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 3.511627906976744,
|
| 793 |
+
"grad_norm": 0.0523432120680809,
|
| 794 |
+
"learning_rate": 4.084577418496775e-06,
|
| 795 |
+
"logits/chosen": -0.20889827609062195,
|
| 796 |
+
"logits/rejected": -0.2736192047595978,
|
| 797 |
+
"logps/chosen": -1008.3775024414062,
|
| 798 |
+
"logps/rejected": -1610.5330810546875,
|
| 799 |
+
"loss": 0.0287,
|
| 800 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 801 |
+
"rewards/chosen": 8.270615577697754,
|
| 802 |
+
"rewards/margins": 36.194480895996094,
|
| 803 |
+
"rewards/rejected": -27.923864364624023,
|
| 804 |
+
"step": 530
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.5780730897009967,
|
| 808 |
+
"grad_norm": 0.014995710924267769,
|
| 809 |
+
"learning_rate": 4.039153688314146e-06,
|
| 810 |
+
"logits/chosen": -0.20946355164051056,
|
| 811 |
+
"logits/rejected": -0.26986002922058105,
|
| 812 |
+
"logps/chosen": -999.2109375,
|
| 813 |
+
"logps/rejected": -1588.408447265625,
|
| 814 |
+
"loss": 0.0329,
|
| 815 |
+
"rewards/accuracies": 0.953125,
|
| 816 |
+
"rewards/chosen": 8.453977584838867,
|
| 817 |
+
"rewards/margins": 36.9298095703125,
|
| 818 |
+
"rewards/rejected": -28.475831985473633,
|
| 819 |
+
"step": 540
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 3.644518272425249,
|
| 823 |
+
"grad_norm": 0.02130577713251114,
|
| 824 |
+
"learning_rate": 3.992896479256966e-06,
|
| 825 |
+
"logits/chosen": -0.21112115681171417,
|
| 826 |
+
"logits/rejected": -0.27220192551612854,
|
| 827 |
+
"logps/chosen": -984.7587890625,
|
| 828 |
+
"logps/rejected": -1543.271240234375,
|
| 829 |
+
"loss": 0.0361,
|
| 830 |
+
"rewards/accuracies": 0.9484374523162842,
|
| 831 |
+
"rewards/chosen": 8.06696891784668,
|
| 832 |
+
"rewards/margins": 37.377586364746094,
|
| 833 |
+
"rewards/rejected": -29.31061553955078,
|
| 834 |
+
"step": 550
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 3.7109634551495017,
|
| 838 |
+
"grad_norm": 0.022987911477684975,
|
| 839 |
+
"learning_rate": 3.945830840419966e-06,
|
| 840 |
+
"logits/chosen": -0.18797051906585693,
|
| 841 |
+
"logits/rejected": -0.2677266299724579,
|
| 842 |
+
"logps/chosen": -1031.5941162109375,
|
| 843 |
+
"logps/rejected": -1641.813720703125,
|
| 844 |
+
"loss": 0.0263,
|
| 845 |
+
"rewards/accuracies": 0.9625000357627869,
|
| 846 |
+
"rewards/chosen": 8.485164642333984,
|
| 847 |
+
"rewards/margins": 39.16246795654297,
|
| 848 |
+
"rewards/rejected": -30.67730140686035,
|
| 849 |
+
"step": 560
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 3.777408637873754,
|
| 853 |
+
"grad_norm": 0.010663657449185848,
|
| 854 |
+
"learning_rate": 3.897982258676867e-06,
|
| 855 |
+
"logits/chosen": -0.19637425243854523,
|
| 856 |
+
"logits/rejected": -0.2760464549064636,
|
| 857 |
+
"logps/chosen": -961.6597900390625,
|
| 858 |
+
"logps/rejected": -1573.847412109375,
|
| 859 |
+
"loss": 0.0383,
|
| 860 |
+
"rewards/accuracies": 0.9453125,
|
| 861 |
+
"rewards/chosen": 8.17518424987793,
|
| 862 |
+
"rewards/margins": 37.773780822753906,
|
| 863 |
+
"rewards/rejected": -29.59859275817871,
|
| 864 |
+
"step": 570
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 3.8438538205980066,
|
| 868 |
+
"grad_norm": 0.05362631008028984,
|
| 869 |
+
"learning_rate": 3.849376644878783e-06,
|
| 870 |
+
"logits/chosen": -0.2064908742904663,
|
| 871 |
+
"logits/rejected": -0.26836133003234863,
|
| 872 |
+
"logps/chosen": -1019.5757446289062,
|
| 873 |
+
"logps/rejected": -1615.7822265625,
|
| 874 |
+
"loss": 0.028,
|
| 875 |
+
"rewards/accuracies": 0.9609374403953552,
|
| 876 |
+
"rewards/chosen": 7.7137041091918945,
|
| 877 |
+
"rewards/margins": 40.03011703491211,
|
| 878 |
+
"rewards/rejected": -32.31641387939453,
|
| 879 |
+
"step": 580
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 3.910299003322259,
|
| 883 |
+
"grad_norm": 0.0628666952252388,
|
| 884 |
+
"learning_rate": 3.8000403198230385e-06,
|
| 885 |
+
"logits/chosen": -0.20015954971313477,
|
| 886 |
+
"logits/rejected": -0.2674901783466339,
|
| 887 |
+
"logps/chosen": -986.7999267578125,
|
| 888 |
+
"logps/rejected": -1530.0762939453125,
|
| 889 |
+
"loss": 0.0223,
|
| 890 |
+
"rewards/accuracies": 0.9703125357627869,
|
| 891 |
+
"rewards/chosen": 8.094964981079102,
|
| 892 |
+
"rewards/margins": 36.99173355102539,
|
| 893 |
+
"rewards/rejected": -28.896766662597656,
|
| 894 |
+
"step": 590
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 3.9767441860465116,
|
| 898 |
+
"grad_norm": 0.27723661065101624,
|
| 899 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 900 |
+
"logits/chosen": -0.21333180367946625,
|
| 901 |
+
"logits/rejected": -0.2772447466850281,
|
| 902 |
+
"logps/chosen": -982.5689086914062,
|
| 903 |
+
"logps/rejected": -1552.4580078125,
|
| 904 |
+
"loss": 0.0357,
|
| 905 |
+
"rewards/accuracies": 0.949999988079071,
|
| 906 |
+
"rewards/chosen": 8.328544616699219,
|
| 907 |
+
"rewards/margins": 38.28776550292969,
|
| 908 |
+
"rewards/rejected": -29.959217071533203,
|
| 909 |
+
"step": 600
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 4.039867109634551,
|
| 913 |
+
"grad_norm": 0.08139760792255402,
|
| 914 |
+
"learning_rate": 3.699282783125616e-06,
|
| 915 |
+
"logits/chosen": -0.20954497158527374,
|
| 916 |
+
"logits/rejected": -0.2789776921272278,
|
| 917 |
+
"logps/chosen": -946.962646484375,
|
| 918 |
+
"logps/rejected": -1561.8204345703125,
|
| 919 |
+
"loss": 0.034,
|
| 920 |
+
"rewards/accuracies": 0.9490132331848145,
|
| 921 |
+
"rewards/chosen": 7.781158447265625,
|
| 922 |
+
"rewards/margins": 38.58467102050781,
|
| 923 |
+
"rewards/rejected": -30.80351448059082,
|
| 924 |
+
"step": 610
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 4.106312292358804,
|
| 928 |
+
"grad_norm": 0.005049778148531914,
|
| 929 |
+
"learning_rate": 3.6479161334675294e-06,
|
| 930 |
+
"logits/chosen": -0.18214014172554016,
|
| 931 |
+
"logits/rejected": -0.27096524834632874,
|
| 932 |
+
"logps/chosen": -959.0311279296875,
|
| 933 |
+
"logps/rejected": -1593.2392578125,
|
| 934 |
+
"loss": 0.0199,
|
| 935 |
+
"rewards/accuracies": 0.971875011920929,
|
| 936 |
+
"rewards/chosen": 8.438522338867188,
|
| 937 |
+
"rewards/margins": 38.53790283203125,
|
| 938 |
+
"rewards/rejected": -30.099374771118164,
|
| 939 |
+
"step": 620
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 4.172757475083056,
|
| 943 |
+
"grad_norm": 0.0067527154460549355,
|
| 944 |
+
"learning_rate": 3.595927866972694e-06,
|
| 945 |
+
"logits/chosen": -0.20282122492790222,
|
| 946 |
+
"logits/rejected": -0.2759528160095215,
|
| 947 |
+
"logps/chosen": -988.6021728515625,
|
| 948 |
+
"logps/rejected": -1591.4195556640625,
|
| 949 |
+
"loss": 0.0251,
|
| 950 |
+
"rewards/accuracies": 0.964062511920929,
|
| 951 |
+
"rewards/chosen": 8.654806137084961,
|
| 952 |
+
"rewards/margins": 39.041419982910156,
|
| 953 |
+
"rewards/rejected": -30.386611938476562,
|
| 954 |
+
"step": 630
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 4.239202657807309,
|
| 958 |
+
"grad_norm": 0.023877006024122238,
|
| 959 |
+
"learning_rate": 3.543346136204545e-06,
|
| 960 |
+
"logits/chosen": -0.20506156980991364,
|
| 961 |
+
"logits/rejected": -0.27975520491600037,
|
| 962 |
+
"logps/chosen": -1013.73681640625,
|
| 963 |
+
"logps/rejected": -1679.869384765625,
|
| 964 |
+
"loss": 0.0339,
|
| 965 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 966 |
+
"rewards/chosen": 8.956099510192871,
|
| 967 |
+
"rewards/margins": 41.54960250854492,
|
| 968 |
+
"rewards/rejected": -32.593502044677734,
|
| 969 |
+
"step": 640
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 4.305647840531561,
|
| 973 |
+
"grad_norm": 0.16767418384552002,
|
| 974 |
+
"learning_rate": 3.4901994150978926e-06,
|
| 975 |
+
"logits/chosen": -0.2239498496055603,
|
| 976 |
+
"logits/rejected": -0.27226123213768005,
|
| 977 |
+
"logps/chosen": -1039.27685546875,
|
| 978 |
+
"logps/rejected": -1601.485107421875,
|
| 979 |
+
"loss": 0.0264,
|
| 980 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 981 |
+
"rewards/chosen": 9.188974380493164,
|
| 982 |
+
"rewards/margins": 39.43864440917969,
|
| 983 |
+
"rewards/rejected": -30.249670028686523,
|
| 984 |
+
"step": 650
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 4.372093023255814,
|
| 988 |
+
"grad_norm": 0.023540904745459557,
|
| 989 |
+
"learning_rate": 3.436516483539781e-06,
|
| 990 |
+
"logits/chosen": -0.21794407069683075,
|
| 991 |
+
"logits/rejected": -0.282469779253006,
|
| 992 |
+
"logps/chosen": -966.4031372070312,
|
| 993 |
+
"logps/rejected": -1527.18310546875,
|
| 994 |
+
"loss": 0.0416,
|
| 995 |
+
"rewards/accuracies": 0.940625011920929,
|
| 996 |
+
"rewards/chosen": 7.827475070953369,
|
| 997 |
+
"rewards/margins": 37.115562438964844,
|
| 998 |
+
"rewards/rejected": -29.2880859375,
|
| 999 |
+
"step": 660
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 4.438538205980066,
|
| 1003 |
+
"grad_norm": 0.0667799860239029,
|
| 1004 |
+
"learning_rate": 3.3823264117846722e-06,
|
| 1005 |
+
"logits/chosen": -0.200710266828537,
|
| 1006 |
+
"logits/rejected": -0.2715788185596466,
|
| 1007 |
+
"logps/chosen": -1030.8763427734375,
|
| 1008 |
+
"logps/rejected": -1679.390869140625,
|
| 1009 |
+
"loss": 0.0243,
|
| 1010 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1011 |
+
"rewards/chosen": 7.525571823120117,
|
| 1012 |
+
"rewards/margins": 43.3552360534668,
|
| 1013 |
+
"rewards/rejected": -35.82966232299805,
|
| 1014 |
+
"step": 670
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 4.504983388704319,
|
| 1018 |
+
"grad_norm": 0.01825464330613613,
|
| 1019 |
+
"learning_rate": 3.3276585447123957e-06,
|
| 1020 |
+
"logits/chosen": -0.19761598110198975,
|
| 1021 |
+
"logits/rejected": -0.2707269787788391,
|
| 1022 |
+
"logps/chosen": -1000.3179931640625,
|
| 1023 |
+
"logps/rejected": -1630.790283203125,
|
| 1024 |
+
"loss": 0.0233,
|
| 1025 |
+
"rewards/accuracies": 0.9671875238418579,
|
| 1026 |
+
"rewards/chosen": 8.050663948059082,
|
| 1027 |
+
"rewards/margins": 41.542236328125,
|
| 1028 |
+
"rewards/rejected": -33.49156951904297,
|
| 1029 |
+
"step": 680
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 4.571428571428571,
|
| 1033 |
+
"grad_norm": 0.010783619247376919,
|
| 1034 |
+
"learning_rate": 3.272542485937369e-06,
|
| 1035 |
+
"logits/chosen": -0.21159793436527252,
|
| 1036 |
+
"logits/rejected": -0.2814417779445648,
|
| 1037 |
+
"logps/chosen": -988.6871337890625,
|
| 1038 |
+
"logps/rejected": -1556.5059814453125,
|
| 1039 |
+
"loss": 0.0295,
|
| 1040 |
+
"rewards/accuracies": 0.9578125476837158,
|
| 1041 |
+
"rewards/chosen": 8.39299201965332,
|
| 1042 |
+
"rewards/margins": 38.407005310058594,
|
| 1043 |
+
"rewards/rejected": -30.01401710510254,
|
| 1044 |
+
"step": 690
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 4.637873754152824,
|
| 1048 |
+
"grad_norm": 0.040845226496458054,
|
| 1049 |
+
"learning_rate": 3.217008081777726e-06,
|
| 1050 |
+
"logits/chosen": -0.2169944941997528,
|
| 1051 |
+
"logits/rejected": -0.28610751032829285,
|
| 1052 |
+
"logps/chosen": -930.8607788085938,
|
| 1053 |
+
"logps/rejected": -1484.1146240234375,
|
| 1054 |
+
"loss": 0.0346,
|
| 1055 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 1056 |
+
"rewards/chosen": 9.505171775817871,
|
| 1057 |
+
"rewards/margins": 35.94363784790039,
|
| 1058 |
+
"rewards/rejected": -26.43846321105957,
|
| 1059 |
+
"step": 700
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 4.704318936877076,
|
| 1063 |
+
"grad_norm": 0.005064092110842466,
|
| 1064 |
+
"learning_rate": 3.1610854050930063e-06,
|
| 1065 |
+
"logits/chosen": -0.1968638151884079,
|
| 1066 |
+
"logits/rejected": -0.274067759513855,
|
| 1067 |
+
"logps/chosen": -988.1194458007812,
|
| 1068 |
+
"logps/rejected": -1650.0118408203125,
|
| 1069 |
+
"loss": 0.0294,
|
| 1070 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1071 |
+
"rewards/chosen": 9.581026077270508,
|
| 1072 |
+
"rewards/margins": 39.75960159301758,
|
| 1073 |
+
"rewards/rejected": -30.17857551574707,
|
| 1074 |
+
"step": 710
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 4.770764119601329,
|
| 1078 |
+
"grad_norm": 0.0075860158540308475,
|
| 1079 |
+
"learning_rate": 3.1048047389991693e-06,
|
| 1080 |
+
"logits/chosen": -0.20798590779304504,
|
| 1081 |
+
"logits/rejected": -0.2726728320121765,
|
| 1082 |
+
"logps/chosen": -1006.69873046875,
|
| 1083 |
+
"logps/rejected": -1619.1817626953125,
|
| 1084 |
+
"loss": 0.0327,
|
| 1085 |
+
"rewards/accuracies": 0.9546875357627869,
|
| 1086 |
+
"rewards/chosen": 8.933056831359863,
|
| 1087 |
+
"rewards/margins": 39.96248245239258,
|
| 1088 |
+
"rewards/rejected": -31.0294246673584,
|
| 1089 |
+
"step": 720
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 4.837209302325581,
|
| 1093 |
+
"grad_norm": 0.018649380654096603,
|
| 1094 |
+
"learning_rate": 3.0481965604697582e-06,
|
| 1095 |
+
"logits/chosen": -0.19381622970104218,
|
| 1096 |
+
"logits/rejected": -0.27504754066467285,
|
| 1097 |
+
"logps/chosen": -964.4519653320312,
|
| 1098 |
+
"logps/rejected": -1575.1983642578125,
|
| 1099 |
+
"loss": 0.024,
|
| 1100 |
+
"rewards/accuracies": 0.9656250476837158,
|
| 1101 |
+
"rewards/chosen": 8.419488906860352,
|
| 1102 |
+
"rewards/margins": 39.957523345947266,
|
| 1103 |
+
"rewards/rejected": -31.53803825378418,
|
| 1104 |
+
"step": 730
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 4.903654485049834,
|
| 1108 |
+
"grad_norm": 0.009096500463783741,
|
| 1109 |
+
"learning_rate": 2.9912915238320755e-06,
|
| 1110 |
+
"logits/chosen": -0.20088475942611694,
|
| 1111 |
+
"logits/rejected": -0.2820194959640503,
|
| 1112 |
+
"logps/chosen": -966.4456787109375,
|
| 1113 |
+
"logps/rejected": -1618.930908203125,
|
| 1114 |
+
"loss": 0.0306,
|
| 1115 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1116 |
+
"rewards/chosen": 9.212898254394531,
|
| 1117 |
+
"rewards/margins": 41.09564208984375,
|
| 1118 |
+
"rewards/rejected": -31.882747650146484,
|
| 1119 |
+
"step": 740
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 4.970099667774086,
|
| 1123 |
+
"grad_norm": 0.01880677044391632,
|
| 1124 |
+
"learning_rate": 2.9341204441673267e-06,
|
| 1125 |
+
"logits/chosen": -0.2191881388425827,
|
| 1126 |
+
"logits/rejected": -0.2813561260700226,
|
| 1127 |
+
"logps/chosen": -1037.1162109375,
|
| 1128 |
+
"logps/rejected": -1705.19873046875,
|
| 1129 |
+
"loss": 0.0305,
|
| 1130 |
+
"rewards/accuracies": 0.9562500715255737,
|
| 1131 |
+
"rewards/chosen": 8.659435272216797,
|
| 1132 |
+
"rewards/margins": 43.9373664855957,
|
| 1133 |
+
"rewards/rejected": -35.27792739868164,
|
| 1134 |
+
"step": 750
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 5.033222591362127,
|
| 1138 |
+
"grad_norm": 0.015276752412319183,
|
| 1139 |
+
"learning_rate": 2.876714280623708e-06,
|
| 1140 |
+
"logits/chosen": -0.1968104988336563,
|
| 1141 |
+
"logits/rejected": -0.27359306812286377,
|
| 1142 |
+
"logps/chosen": -971.7843017578125,
|
| 1143 |
+
"logps/rejected": -1637.6700439453125,
|
| 1144 |
+
"loss": 0.023,
|
| 1145 |
+
"rewards/accuracies": 0.9654605388641357,
|
| 1146 |
+
"rewards/chosen": 8.207253456115723,
|
| 1147 |
+
"rewards/margins": 42.495601654052734,
|
| 1148 |
+
"rewards/rejected": -34.28834533691406,
|
| 1149 |
+
"step": 760
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 5.099667774086379,
|
| 1153 |
+
"grad_norm": 0.005275533068925142,
|
| 1154 |
+
"learning_rate": 2.8191041196514874e-06,
|
| 1155 |
+
"logits/chosen": -0.198197603225708,
|
| 1156 |
+
"logits/rejected": -0.28718599677085876,
|
| 1157 |
+
"logps/chosen": -1012.2830200195312,
|
| 1158 |
+
"logps/rejected": -1695.5068359375,
|
| 1159 |
+
"loss": 0.0359,
|
| 1160 |
+
"rewards/accuracies": 0.9484374523162842,
|
| 1161 |
+
"rewards/chosen": 8.050834655761719,
|
| 1162 |
+
"rewards/margins": 43.739540100097656,
|
| 1163 |
+
"rewards/rejected": -35.688697814941406,
|
| 1164 |
+
"step": 770
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 5.166112956810632,
|
| 1168 |
+
"grad_norm": 0.03785371407866478,
|
| 1169 |
+
"learning_rate": 2.761321158169134e-06,
|
| 1170 |
+
"logits/chosen": -0.21472084522247314,
|
| 1171 |
+
"logits/rejected": -0.2798621654510498,
|
| 1172 |
+
"logps/chosen": -1062.145263671875,
|
| 1173 |
+
"logps/rejected": -1711.7852783203125,
|
| 1174 |
+
"loss": 0.0338,
|
| 1175 |
+
"rewards/accuracies": 0.953125,
|
| 1176 |
+
"rewards/chosen": 8.760993003845215,
|
| 1177 |
+
"rewards/margins": 44.73903274536133,
|
| 1178 |
+
"rewards/rejected": -35.9780387878418,
|
| 1179 |
+
"step": 780
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 5.232558139534884,
|
| 1183 |
+
"grad_norm": 0.05422116443514824,
|
| 1184 |
+
"learning_rate": 2.703396686669646e-06,
|
| 1185 |
+
"logits/chosen": -0.2090332806110382,
|
| 1186 |
+
"logits/rejected": -0.2846227288246155,
|
| 1187 |
+
"logps/chosen": -984.9669189453125,
|
| 1188 |
+
"logps/rejected": -1620.4434814453125,
|
| 1189 |
+
"loss": 0.0338,
|
| 1190 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 1191 |
+
"rewards/chosen": 8.526132583618164,
|
| 1192 |
+
"rewards/margins": 41.130706787109375,
|
| 1193 |
+
"rewards/rejected": -32.60457229614258,
|
| 1194 |
+
"step": 790
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 5.299003322259137,
|
| 1198 |
+
"grad_norm": 0.021388115361332893,
|
| 1199 |
+
"learning_rate": 2.6453620722761897e-06,
|
| 1200 |
+
"logits/chosen": -0.22062420845031738,
|
| 1201 |
+
"logits/rejected": -0.2795643210411072,
|
| 1202 |
+
"logps/chosen": -1018.5145263671875,
|
| 1203 |
+
"logps/rejected": -1654.4093017578125,
|
| 1204 |
+
"loss": 0.0305,
|
| 1205 |
+
"rewards/accuracies": 0.9562499523162842,
|
| 1206 |
+
"rewards/chosen": 9.038843154907227,
|
| 1207 |
+
"rewards/margins": 43.468109130859375,
|
| 1208 |
+
"rewards/rejected": -34.42927169799805,
|
| 1209 |
+
"step": 800
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 5.365448504983389,
|
| 1213 |
+
"grad_norm": 0.02929343283176422,
|
| 1214 |
+
"learning_rate": 2.587248741756253e-06,
|
| 1215 |
+
"logits/chosen": -0.19668123126029968,
|
| 1216 |
+
"logits/rejected": -0.2809968888759613,
|
| 1217 |
+
"logps/chosen": -1002.3245239257812,
|
| 1218 |
+
"logps/rejected": -1654.1307373046875,
|
| 1219 |
+
"loss": 0.0203,
|
| 1220 |
+
"rewards/accuracies": 0.9718750715255737,
|
| 1221 |
+
"rewards/chosen": 7.798498630523682,
|
| 1222 |
+
"rewards/margins": 43.116241455078125,
|
| 1223 |
+
"rewards/rejected": -35.3177490234375,
|
| 1224 |
+
"step": 810
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 5.431893687707642,
|
| 1228 |
+
"grad_norm": 0.007884162478148937,
|
| 1229 |
+
"learning_rate": 2.5290881645034932e-06,
|
| 1230 |
+
"logits/chosen": -0.2037358433008194,
|
| 1231 |
+
"logits/rejected": -0.2698831558227539,
|
| 1232 |
+
"logps/chosen": -986.9371337890625,
|
| 1233 |
+
"logps/rejected": -1607.3692626953125,
|
| 1234 |
+
"loss": 0.0197,
|
| 1235 |
+
"rewards/accuracies": 0.9734375476837158,
|
| 1236 |
+
"rewards/chosen": 7.649607181549072,
|
| 1237 |
+
"rewards/margins": 42.175437927246094,
|
| 1238 |
+
"rewards/rejected": -34.52583312988281,
|
| 1239 |
+
"step": 820
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 5.498338870431894,
|
| 1243 |
+
"grad_norm": 0.01341085135936737,
|
| 1244 |
+
"learning_rate": 2.470911835496508e-06,
|
| 1245 |
+
"logits/chosen": -0.19817571341991425,
|
| 1246 |
+
"logits/rejected": -0.27480000257492065,
|
| 1247 |
+
"logps/chosen": -1008.7352905273438,
|
| 1248 |
+
"logps/rejected": -1691.955322265625,
|
| 1249 |
+
"loss": 0.024,
|
| 1250 |
+
"rewards/accuracies": 0.9656250476837158,
|
| 1251 |
+
"rewards/chosen": 7.666060447692871,
|
| 1252 |
+
"rewards/margins": 44.92629623413086,
|
| 1253 |
+
"rewards/rejected": -37.26023864746094,
|
| 1254 |
+
"step": 830
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 5.5647840531561465,
|
| 1258 |
+
"grad_norm": 0.19932714104652405,
|
| 1259 |
+
"learning_rate": 2.4127512582437486e-06,
|
| 1260 |
+
"logits/chosen": -0.21822643280029297,
|
| 1261 |
+
"logits/rejected": -0.2849328815937042,
|
| 1262 |
+
"logps/chosen": -952.5464477539062,
|
| 1263 |
+
"logps/rejected": -1569.07470703125,
|
| 1264 |
+
"loss": 0.0351,
|
| 1265 |
+
"rewards/accuracies": 0.9500000476837158,
|
| 1266 |
+
"rewards/chosen": 8.396932601928711,
|
| 1267 |
+
"rewards/margins": 40.544769287109375,
|
| 1268 |
+
"rewards/rejected": -32.14783477783203,
|
| 1269 |
+
"step": 840
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 5.631229235880399,
|
| 1273 |
+
"grad_norm": 0.024993257597088814,
|
| 1274 |
+
"learning_rate": 2.3546379277238107e-06,
|
| 1275 |
+
"logits/chosen": -0.2073173075914383,
|
| 1276 |
+
"logits/rejected": -0.2831364870071411,
|
| 1277 |
+
"logps/chosen": -982.529052734375,
|
| 1278 |
+
"logps/rejected": -1603.887939453125,
|
| 1279 |
+
"loss": 0.023,
|
| 1280 |
+
"rewards/accuracies": 0.9671875238418579,
|
| 1281 |
+
"rewards/chosen": 9.977124214172363,
|
| 1282 |
+
"rewards/margins": 41.838199615478516,
|
| 1283 |
+
"rewards/rejected": -31.8610782623291,
|
| 1284 |
+
"step": 850
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 5.6976744186046515,
|
| 1288 |
+
"grad_norm": 0.025327615439891815,
|
| 1289 |
+
"learning_rate": 2.296603313330355e-06,
|
| 1290 |
+
"logits/chosen": -0.22604137659072876,
|
| 1291 |
+
"logits/rejected": -0.2886336147785187,
|
| 1292 |
+
"logps/chosen": -949.3585205078125,
|
| 1293 |
+
"logps/rejected": -1545.9788818359375,
|
| 1294 |
+
"loss": 0.0393,
|
| 1295 |
+
"rewards/accuracies": 0.9437499642372131,
|
| 1296 |
+
"rewards/chosen": 9.380352973937988,
|
| 1297 |
+
"rewards/margins": 39.31431198120117,
|
| 1298 |
+
"rewards/rejected": -29.9339599609375,
|
| 1299 |
+
"step": 860
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 5.764119601328904,
|
| 1303 |
+
"grad_norm": 0.09348896890878677,
|
| 1304 |
+
"learning_rate": 2.238678841830867e-06,
|
| 1305 |
+
"logits/chosen": -0.21441370248794556,
|
| 1306 |
+
"logits/rejected": -0.28145626187324524,
|
| 1307 |
+
"logps/chosen": -969.75927734375,
|
| 1308 |
+
"logps/rejected": -1589.4075927734375,
|
| 1309 |
+
"loss": 0.0326,
|
| 1310 |
+
"rewards/accuracies": 0.9531249403953552,
|
| 1311 |
+
"rewards/chosen": 8.257079124450684,
|
| 1312 |
+
"rewards/margins": 40.917396545410156,
|
| 1313 |
+
"rewards/rejected": -32.660316467285156,
|
| 1314 |
+
"step": 870
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 5.8305647840531565,
|
| 1318 |
+
"grad_norm": 0.04749366641044617,
|
| 1319 |
+
"learning_rate": 2.1808958803485134e-06,
|
| 1320 |
+
"logits/chosen": -0.20293860137462616,
|
| 1321 |
+
"logits/rejected": -0.2881726026535034,
|
| 1322 |
+
"logps/chosen": -956.5830078125,
|
| 1323 |
+
"logps/rejected": -1648.4208984375,
|
| 1324 |
+
"loss": 0.0294,
|
| 1325 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1326 |
+
"rewards/chosen": 8.458630561828613,
|
| 1327 |
+
"rewards/margins": 43.482696533203125,
|
| 1328 |
+
"rewards/rejected": -35.02406311035156,
|
| 1329 |
+
"step": 880
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 5.897009966777409,
|
| 1333 |
+
"grad_norm": 0.016179274767637253,
|
| 1334 |
+
"learning_rate": 2.1232857193762923e-06,
|
| 1335 |
+
"logits/chosen": -0.2071429193019867,
|
| 1336 |
+
"logits/rejected": -0.2786925435066223,
|
| 1337 |
+
"logps/chosen": -974.3377685546875,
|
| 1338 |
+
"logps/rejected": -1581.3140869140625,
|
| 1339 |
+
"loss": 0.0284,
|
| 1340 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 1341 |
+
"rewards/chosen": 8.533307075500488,
|
| 1342 |
+
"rewards/margins": 41.709964752197266,
|
| 1343 |
+
"rewards/rejected": -33.176658630371094,
|
| 1344 |
+
"step": 890
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 5.9634551495016614,
|
| 1348 |
+
"grad_norm": 0.0812649205327034,
|
| 1349 |
+
"learning_rate": 2.0658795558326745e-06,
|
| 1350 |
+
"logits/chosen": -0.21706193685531616,
|
| 1351 |
+
"logits/rejected": -0.2818005681037903,
|
| 1352 |
+
"logps/chosen": -990.1275634765625,
|
| 1353 |
+
"logps/rejected": -1619.1646728515625,
|
| 1354 |
+
"loss": 0.0316,
|
| 1355 |
+
"rewards/accuracies": 0.9546875357627869,
|
| 1356 |
+
"rewards/chosen": 8.88737678527832,
|
| 1357 |
+
"rewards/margins": 41.76374053955078,
|
| 1358 |
+
"rewards/rejected": -32.876365661621094,
|
| 1359 |
+
"step": 900
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 6.026578073089701,
|
| 1363 |
+
"grad_norm": 0.009111125022172928,
|
| 1364 |
+
"learning_rate": 2.0087084761679245e-06,
|
| 1365 |
+
"logits/chosen": -0.2077750265598297,
|
| 1366 |
+
"logits/rejected": -0.2753692865371704,
|
| 1367 |
+
"logps/chosen": -1017.2333984375,
|
| 1368 |
+
"logps/rejected": -1682.0185546875,
|
| 1369 |
+
"loss": 0.0176,
|
| 1370 |
+
"rewards/accuracies": 0.9736842513084412,
|
| 1371 |
+
"rewards/chosen": 8.799712181091309,
|
| 1372 |
+
"rewards/margins": 43.73455810546875,
|
| 1373 |
+
"rewards/rejected": -34.934844970703125,
|
| 1374 |
+
"step": 910
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 6.093023255813954,
|
| 1378 |
+
"grad_norm": 0.009804153814911842,
|
| 1379 |
+
"learning_rate": 1.9518034395302413e-06,
|
| 1380 |
+
"logits/chosen": -0.19864043593406677,
|
| 1381 |
+
"logits/rejected": -0.2817736566066742,
|
| 1382 |
+
"logps/chosen": -986.0798950195312,
|
| 1383 |
+
"logps/rejected": -1684.548095703125,
|
| 1384 |
+
"loss": 0.0274,
|
| 1385 |
+
"rewards/accuracies": 0.9609375596046448,
|
| 1386 |
+
"rewards/chosen": 9.566544532775879,
|
| 1387 |
+
"rewards/margins": 43.709625244140625,
|
| 1388 |
+
"rewards/rejected": -34.14308547973633,
|
| 1389 |
+
"step": 920
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 6.159468438538206,
|
| 1393 |
+
"grad_norm": 0.0946379005908966,
|
| 1394 |
+
"learning_rate": 1.895195261000831e-06,
|
| 1395 |
+
"logits/chosen": -0.20698292553424835,
|
| 1396 |
+
"logits/rejected": -0.2797500193119049,
|
| 1397 |
+
"logps/chosen": -999.882568359375,
|
| 1398 |
+
"logps/rejected": -1666.398193359375,
|
| 1399 |
+
"loss": 0.0261,
|
| 1400 |
+
"rewards/accuracies": 0.9625000357627869,
|
| 1401 |
+
"rewards/chosen": 9.238912582397461,
|
| 1402 |
+
"rewards/margins": 43.197425842285156,
|
| 1403 |
+
"rewards/rejected": -33.9585075378418,
|
| 1404 |
+
"step": 930
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 6.225913621262459,
|
| 1408 |
+
"grad_norm": 0.03615960851311684,
|
| 1409 |
+
"learning_rate": 1.8389145949069953e-06,
|
| 1410 |
+
"logits/chosen": -0.20727947354316711,
|
| 1411 |
+
"logits/rejected": -0.2856772541999817,
|
| 1412 |
+
"logps/chosen": -1012.0263061523438,
|
| 1413 |
+
"logps/rejected": -1656.6219482421875,
|
| 1414 |
+
"loss": 0.0299,
|
| 1415 |
+
"rewards/accuracies": 0.9578125476837158,
|
| 1416 |
+
"rewards/chosen": 9.405717849731445,
|
| 1417 |
+
"rewards/margins": 43.6558952331543,
|
| 1418 |
+
"rewards/rejected": -34.25017547607422,
|
| 1419 |
+
"step": 940
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 6.292358803986711,
|
| 1423 |
+
"grad_norm": 0.09421185404062271,
|
| 1424 |
+
"learning_rate": 1.7829919182222752e-06,
|
| 1425 |
+
"logits/chosen": -0.21414652466773987,
|
| 1426 |
+
"logits/rejected": -0.2791929244995117,
|
| 1427 |
+
"logps/chosen": -991.490966796875,
|
| 1428 |
+
"logps/rejected": -1588.97412109375,
|
| 1429 |
+
"loss": 0.0284,
|
| 1430 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 1431 |
+
"rewards/chosen": 8.861186981201172,
|
| 1432 |
+
"rewards/margins": 42.22901153564453,
|
| 1433 |
+
"rewards/rejected": -33.367828369140625,
|
| 1434 |
+
"step": 950
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 6.358803986710964,
|
| 1438 |
+
"grad_norm": 0.1068965345621109,
|
| 1439 |
+
"learning_rate": 1.7274575140626318e-06,
|
| 1440 |
+
"logits/chosen": -0.20373263955116272,
|
| 1441 |
+
"logits/rejected": -0.2882729768753052,
|
| 1442 |
+
"logps/chosen": -975.765625,
|
| 1443 |
+
"logps/rejected": -1631.05810546875,
|
| 1444 |
+
"loss": 0.0327,
|
| 1445 |
+
"rewards/accuracies": 0.9546874761581421,
|
| 1446 |
+
"rewards/chosen": 9.140608787536621,
|
| 1447 |
+
"rewards/margins": 42.56184387207031,
|
| 1448 |
+
"rewards/rejected": -33.421234130859375,
|
| 1449 |
+
"step": 960
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 6.425249169435216,
|
| 1453 |
+
"grad_norm": 0.20864491164684296,
|
| 1454 |
+
"learning_rate": 1.6723414552876052e-06,
|
| 1455 |
+
"logits/chosen": -0.2219492346048355,
|
| 1456 |
+
"logits/rejected": -0.2909078299999237,
|
| 1457 |
+
"logps/chosen": -998.4899291992188,
|
| 1458 |
+
"logps/rejected": -1629.822998046875,
|
| 1459 |
+
"loss": 0.0316,
|
| 1460 |
+
"rewards/accuracies": 0.9546874761581421,
|
| 1461 |
+
"rewards/chosen": 10.093676567077637,
|
| 1462 |
+
"rewards/margins": 41.07781982421875,
|
| 1463 |
+
"rewards/rejected": -30.98414421081543,
|
| 1464 |
+
"step": 970
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 6.4916943521594686,
|
| 1468 |
+
"grad_norm": 0.0074838376604020596,
|
| 1469 |
+
"learning_rate": 1.6176735882153284e-06,
|
| 1470 |
+
"logits/chosen": -0.21261698007583618,
|
| 1471 |
+
"logits/rejected": -0.2835195064544678,
|
| 1472 |
+
"logps/chosen": -982.4485473632812,
|
| 1473 |
+
"logps/rejected": -1603.0489501953125,
|
| 1474 |
+
"loss": 0.0274,
|
| 1475 |
+
"rewards/accuracies": 0.9609375,
|
| 1476 |
+
"rewards/chosen": 9.805879592895508,
|
| 1477 |
+
"rewards/margins": 41.88581466674805,
|
| 1478 |
+
"rewards/rejected": -32.079933166503906,
|
| 1479 |
+
"step": 980
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 6.558139534883721,
|
| 1483 |
+
"grad_norm": 0.009234702214598656,
|
| 1484 |
+
"learning_rate": 1.56348351646022e-06,
|
| 1485 |
+
"logits/chosen": -0.21804842352867126,
|
| 1486 |
+
"logits/rejected": -0.2872644066810608,
|
| 1487 |
+
"logps/chosen": -992.9108276367188,
|
| 1488 |
+
"logps/rejected": -1682.7227783203125,
|
| 1489 |
+
"loss": 0.0326,
|
| 1490 |
+
"rewards/accuracies": 0.953125,
|
| 1491 |
+
"rewards/chosen": 9.05055046081543,
|
| 1492 |
+
"rewards/margins": 44.20970916748047,
|
| 1493 |
+
"rewards/rejected": -35.15916442871094,
|
| 1494 |
+
"step": 990
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 6.6245847176079735,
|
| 1498 |
+
"grad_norm": 0.005751634947955608,
|
| 1499 |
+
"learning_rate": 1.509800584902108e-06,
|
| 1500 |
+
"logits/chosen": -0.21074561774730682,
|
| 1501 |
+
"logits/rejected": -0.28109174966812134,
|
| 1502 |
+
"logps/chosen": -962.9151611328125,
|
| 1503 |
+
"logps/rejected": -1613.1900634765625,
|
| 1504 |
+
"loss": 0.025,
|
| 1505 |
+
"rewards/accuracies": 0.9640624523162842,
|
| 1506 |
+
"rewards/chosen": 8.966195106506348,
|
| 1507 |
+
"rewards/margins": 42.510337829589844,
|
| 1508 |
+
"rewards/rejected": -33.54414367675781,
|
| 1509 |
+
"step": 1000
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 6.691029900332226,
|
| 1513 |
+
"grad_norm": 0.004595512058585882,
|
| 1514 |
+
"learning_rate": 1.4566538637954556e-06,
|
| 1515 |
+
"logits/chosen": -0.206181600689888,
|
| 1516 |
+
"logits/rejected": -0.28721800446510315,
|
| 1517 |
+
"logps/chosen": -916.551025390625,
|
| 1518 |
+
"logps/rejected": -1526.8843994140625,
|
| 1519 |
+
"loss": 0.023,
|
| 1520 |
+
"rewards/accuracies": 0.9671874642372131,
|
| 1521 |
+
"rewards/chosen": 9.025623321533203,
|
| 1522 |
+
"rewards/margins": 39.80775833129883,
|
| 1523 |
+
"rewards/rejected": -30.782133102416992,
|
| 1524 |
+
"step": 1010
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 6.7574750830564785,
|
| 1528 |
+
"grad_norm": 0.003542843274772167,
|
| 1529 |
+
"learning_rate": 1.4040721330273063e-06,
|
| 1530 |
+
"logits/chosen": -0.2257394641637802,
|
| 1531 |
+
"logits/rejected": -0.2908530533313751,
|
| 1532 |
+
"logps/chosen": -960.90234375,
|
| 1533 |
+
"logps/rejected": -1578.2750244140625,
|
| 1534 |
+
"loss": 0.0434,
|
| 1535 |
+
"rewards/accuracies": 0.9375,
|
| 1536 |
+
"rewards/chosen": 8.649526596069336,
|
| 1537 |
+
"rewards/margins": 42.02993392944336,
|
| 1538 |
+
"rewards/rejected": -33.380401611328125,
|
| 1539 |
+
"step": 1020
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 6.823920265780731,
|
| 1543 |
+
"grad_norm": 0.06329812854528427,
|
| 1544 |
+
"learning_rate": 1.3520838665324704e-06,
|
| 1545 |
+
"logits/chosen": -0.2071303427219391,
|
| 1546 |
+
"logits/rejected": -0.2824835181236267,
|
| 1547 |
+
"logps/chosen": -973.2369995117188,
|
| 1548 |
+
"logps/rejected": -1608.9039306640625,
|
| 1549 |
+
"loss": 0.0295,
|
| 1550 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1551 |
+
"rewards/chosen": 8.745044708251953,
|
| 1552 |
+
"rewards/margins": 42.52543258666992,
|
| 1553 |
+
"rewards/rejected": -33.78038787841797,
|
| 1554 |
+
"step": 1030
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 6.8903654485049834,
|
| 1558 |
+
"grad_norm": 0.014581708237528801,
|
| 1559 |
+
"learning_rate": 1.3007172168743854e-06,
|
| 1560 |
+
"logits/chosen": -0.21656513214111328,
|
| 1561 |
+
"logits/rejected": -0.2857995927333832,
|
| 1562 |
+
"logps/chosen": -993.361328125,
|
| 1563 |
+
"logps/rejected": -1631.5338134765625,
|
| 1564 |
+
"loss": 0.0273,
|
| 1565 |
+
"rewards/accuracies": 0.9609375,
|
| 1566 |
+
"rewards/chosen": 9.786979675292969,
|
| 1567 |
+
"rewards/margins": 42.11618423461914,
|
| 1568 |
+
"rewards/rejected": -32.32920455932617,
|
| 1569 |
+
"step": 1040
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 6.956810631229236,
|
| 1573 |
+
"grad_norm": 0.042626578360795975,
|
| 1574 |
+
"learning_rate": 1.2500000000000007e-06,
|
| 1575 |
+
"logits/chosen": -0.20040905475616455,
|
| 1576 |
+
"logits/rejected": -0.2804938852787018,
|
| 1577 |
+
"logps/chosen": -1012.1871337890625,
|
| 1578 |
+
"logps/rejected": -1693.8763427734375,
|
| 1579 |
+
"loss": 0.024,
|
| 1580 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1581 |
+
"rewards/chosen": 9.336542129516602,
|
| 1582 |
+
"rewards/margins": 44.35811996459961,
|
| 1583 |
+
"rewards/rejected": -35.02157211303711,
|
| 1584 |
+
"step": 1050
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 7.019933554817276,
|
| 1588 |
+
"grad_norm": 0.08022702485322952,
|
| 1589 |
+
"learning_rate": 1.1999596801769617e-06,
|
| 1590 |
+
"logits/chosen": -0.237129807472229,
|
| 1591 |
+
"logits/rejected": -0.28911346197128296,
|
| 1592 |
+
"logps/chosen": -1007.3953857421875,
|
| 1593 |
+
"logps/rejected": -1635.5213623046875,
|
| 1594 |
+
"loss": 0.0337,
|
| 1595 |
+
"rewards/accuracies": 0.9490132331848145,
|
| 1596 |
+
"rewards/chosen": 9.52566909790039,
|
| 1597 |
+
"rewards/margins": 42.855995178222656,
|
| 1598 |
+
"rewards/rejected": -33.33032989501953,
|
| 1599 |
+
"step": 1060
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 7.086378737541528,
|
| 1603 |
+
"grad_norm": 0.022050941362977028,
|
| 1604 |
+
"learning_rate": 1.1506233551212186e-06,
|
| 1605 |
+
"logits/chosen": -0.21040520071983337,
|
| 1606 |
+
"logits/rejected": -0.28766316175460815,
|
| 1607 |
+
"logps/chosen": -1012.5867919921875,
|
| 1608 |
+
"logps/rejected": -1669.2762451171875,
|
| 1609 |
+
"loss": 0.024,
|
| 1610 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1611 |
+
"rewards/chosen": 9.755768775939941,
|
| 1612 |
+
"rewards/margins": 43.82841110229492,
|
| 1613 |
+
"rewards/rejected": -34.07263946533203,
|
| 1614 |
+
"step": 1070
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 7.152823920265781,
|
| 1618 |
+
"grad_norm": 0.006165655329823494,
|
| 1619 |
+
"learning_rate": 1.1020177413231334e-06,
|
| 1620 |
+
"logits/chosen": -0.20854339003562927,
|
| 1621 |
+
"logits/rejected": -0.28692665696144104,
|
| 1622 |
+
"logps/chosen": -970.487060546875,
|
| 1623 |
+
"logps/rejected": -1616.68212890625,
|
| 1624 |
+
"loss": 0.0304,
|
| 1625 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1626 |
+
"rewards/chosen": 9.218266487121582,
|
| 1627 |
+
"rewards/margins": 42.4344367980957,
|
| 1628 |
+
"rewards/rejected": -33.21616744995117,
|
| 1629 |
+
"step": 1080
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 7.219269102990033,
|
| 1633 |
+
"grad_norm": 0.02192852459847927,
|
| 1634 |
+
"learning_rate": 1.0541691595800338e-06,
|
| 1635 |
+
"logits/chosen": -0.20376423001289368,
|
| 1636 |
+
"logits/rejected": -0.2804113030433655,
|
| 1637 |
+
"logps/chosen": -1000.4962158203125,
|
| 1638 |
+
"logps/rejected": -1672.725341796875,
|
| 1639 |
+
"loss": 0.0349,
|
| 1640 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1641 |
+
"rewards/chosen": 8.690274238586426,
|
| 1642 |
+
"rewards/margins": 44.81369400024414,
|
| 1643 |
+
"rewards/rejected": -36.1234130859375,
|
| 1644 |
+
"step": 1090
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 7.285714285714286,
|
| 1648 |
+
"grad_norm": 0.07914319634437561,
|
| 1649 |
+
"learning_rate": 1.0071035207430352e-06,
|
| 1650 |
+
"logits/chosen": -0.19936969876289368,
|
| 1651 |
+
"logits/rejected": -0.28211313486099243,
|
| 1652 |
+
"logps/chosen": -978.279052734375,
|
| 1653 |
+
"logps/rejected": -1641.731201171875,
|
| 1654 |
+
"loss": 0.0262,
|
| 1655 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1656 |
+
"rewards/chosen": 8.959029197692871,
|
| 1657 |
+
"rewards/margins": 43.952213287353516,
|
| 1658 |
+
"rewards/rejected": -34.993186950683594,
|
| 1659 |
+
"step": 1100
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 7.352159468438538,
|
| 1663 |
+
"grad_norm": 0.014706037007272243,
|
| 1664 |
+
"learning_rate": 9.608463116858544e-07,
|
| 1665 |
+
"logits/chosen": -0.21040727198123932,
|
| 1666 |
+
"logits/rejected": -0.2795957922935486,
|
| 1667 |
+
"logps/chosen": -1030.148681640625,
|
| 1668 |
+
"logps/rejected": -1709.89892578125,
|
| 1669 |
+
"loss": 0.0217,
|
| 1670 |
+
"rewards/accuracies": 0.96875,
|
| 1671 |
+
"rewards/chosen": 9.775686264038086,
|
| 1672 |
+
"rewards/margins": 45.375675201416016,
|
| 1673 |
+
"rewards/rejected": -35.59999084472656,
|
| 1674 |
+
"step": 1110
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 7.4186046511627906,
|
| 1678 |
+
"grad_norm": 0.006406854372471571,
|
| 1679 |
+
"learning_rate": 9.154225815032242e-07,
|
| 1680 |
+
"logits/chosen": -0.20644372701644897,
|
| 1681 |
+
"logits/rejected": -0.2845055162906647,
|
| 1682 |
+
"logps/chosen": -941.191162109375,
|
| 1683 |
+
"logps/rejected": -1556.8076171875,
|
| 1684 |
+
"loss": 0.0284,
|
| 1685 |
+
"rewards/accuracies": 0.9593749642372131,
|
| 1686 |
+
"rewards/chosen": 9.463266372680664,
|
| 1687 |
+
"rewards/margins": 41.32258224487305,
|
| 1688 |
+
"rewards/rejected": -31.859315872192383,
|
| 1689 |
+
"step": 1120
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 7.485049833887043,
|
| 1693 |
+
"grad_norm": 0.023300213739275932,
|
| 1694 |
+
"learning_rate": 8.708569279463622e-07,
|
| 1695 |
+
"logits/chosen": -0.2077961415052414,
|
| 1696 |
+
"logits/rejected": -0.27984070777893066,
|
| 1697 |
+
"logps/chosen": -962.038330078125,
|
| 1698 |
+
"logps/rejected": -1615.148193359375,
|
| 1699 |
+
"loss": 0.0273,
|
| 1700 |
+
"rewards/accuracies": 0.9609375,
|
| 1701 |
+
"rewards/chosen": 9.014985084533691,
|
| 1702 |
+
"rewards/margins": 42.5955924987793,
|
| 1703 |
+
"rewards/rejected": -33.580604553222656,
|
| 1704 |
+
"step": 1130
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 7.5514950166112955,
|
| 1708 |
+
"grad_norm": 0.0013064617523923516,
|
| 1709 |
+
"learning_rate": 8.271734841028553e-07,
|
| 1710 |
+
"logits/chosen": -0.21521812677383423,
|
| 1711 |
+
"logits/rejected": -0.29194578528404236,
|
| 1712 |
+
"logps/chosen": -969.0303955078125,
|
| 1713 |
+
"logps/rejected": -1659.9932861328125,
|
| 1714 |
+
"loss": 0.0305,
|
| 1715 |
+
"rewards/accuracies": 0.9562500715255737,
|
| 1716 |
+
"rewards/chosen": 9.51773452758789,
|
| 1717 |
+
"rewards/margins": 43.94035720825195,
|
| 1718 |
+
"rewards/rejected": -34.42262268066406,
|
| 1719 |
+
"step": 1140
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 7.617940199335548,
|
| 1723 |
+
"grad_norm": 0.027837639674544334,
|
| 1724 |
+
"learning_rate": 7.843959053281663e-07,
|
| 1725 |
+
"logits/chosen": -0.2188502699136734,
|
| 1726 |
+
"logits/rejected": -0.28792399168014526,
|
| 1727 |
+
"logps/chosen": -983.6058959960938,
|
| 1728 |
+
"logps/rejected": -1592.7557373046875,
|
| 1729 |
+
"loss": 0.024,
|
| 1730 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1731 |
+
"rewards/chosen": 9.258206367492676,
|
| 1732 |
+
"rewards/margins": 42.35002136230469,
|
| 1733 |
+
"rewards/rejected": -33.09181213378906,
|
| 1734 |
+
"step": 1150
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 7.6843853820598005,
|
| 1738 |
+
"grad_norm": 0.021290810778737068,
|
| 1739 |
+
"learning_rate": 7.425473564358457e-07,
|
| 1740 |
+
"logits/chosen": -0.21639959514141083,
|
| 1741 |
+
"logits/rejected": -0.2931935489177704,
|
| 1742 |
+
"logps/chosen": -947.8458251953125,
|
| 1743 |
+
"logps/rejected": -1650.803466796875,
|
| 1744 |
+
"loss": 0.0348,
|
| 1745 |
+
"rewards/accuracies": 0.9499999284744263,
|
| 1746 |
+
"rewards/chosen": 9.148832321166992,
|
| 1747 |
+
"rewards/margins": 43.69053268432617,
|
| 1748 |
+
"rewards/rejected": -34.54170227050781,
|
| 1749 |
+
"step": 1160
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 7.750830564784053,
|
| 1753 |
+
"grad_norm": 0.0003092914412263781,
|
| 1754 |
+
"learning_rate": 7.016504991533727e-07,
|
| 1755 |
+
"logits/chosen": -0.21218986809253693,
|
| 1756 |
+
"logits/rejected": -0.2906990051269531,
|
| 1757 |
+
"logps/chosen": -1019.6378173828125,
|
| 1758 |
+
"logps/rejected": -1668.9599609375,
|
| 1759 |
+
"loss": 0.0273,
|
| 1760 |
+
"rewards/accuracies": 0.9609375,
|
| 1761 |
+
"rewards/chosen": 9.230569839477539,
|
| 1762 |
+
"rewards/margins": 43.63245391845703,
|
| 1763 |
+
"rewards/rejected": -34.40188217163086,
|
| 1764 |
+
"step": 1170
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 7.8172757475083055,
|
| 1768 |
+
"grad_norm": 0.06043156608939171,
|
| 1769 |
+
"learning_rate": 6.617274798504286e-07,
|
| 1770 |
+
"logits/chosen": -0.22491909563541412,
|
| 1771 |
+
"logits/rejected": -0.287073016166687,
|
| 1772 |
+
"logps/chosen": -969.3558349609375,
|
| 1773 |
+
"logps/rejected": -1639.527099609375,
|
| 1774 |
+
"loss": 0.038,
|
| 1775 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 1776 |
+
"rewards/chosen": 9.15910530090332,
|
| 1777 |
+
"rewards/margins": 43.614559173583984,
|
| 1778 |
+
"rewards/rejected": -34.45545196533203,
|
| 1779 |
+
"step": 1180
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 7.883720930232558,
|
| 1783 |
+
"grad_norm": 0.07719282060861588,
|
| 1784 |
+
"learning_rate": 6.227999175462521e-07,
|
| 1785 |
+
"logits/chosen": -0.21083196997642517,
|
| 1786 |
+
"logits/rejected": -0.2891797125339508,
|
| 1787 |
+
"logps/chosen": -949.5618286132812,
|
| 1788 |
+
"logps/rejected": -1556.614990234375,
|
| 1789 |
+
"loss": 0.03,
|
| 1790 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1791 |
+
"rewards/chosen": 9.331879615783691,
|
| 1792 |
+
"rewards/margins": 40.77275466918945,
|
| 1793 |
+
"rewards/rejected": -31.440876007080078,
|
| 1794 |
+
"step": 1190
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 7.95016611295681,
|
| 1798 |
+
"grad_norm": 0.0071638901717960835,
|
| 1799 |
+
"learning_rate": 5.848888922025553e-07,
|
| 1800 |
+
"logits/chosen": -0.21286137402057648,
|
| 1801 |
+
"logits/rejected": -0.2848753035068512,
|
| 1802 |
+
"logps/chosen": -994.6892700195312,
|
| 1803 |
+
"logps/rejected": -1621.3260498046875,
|
| 1804 |
+
"loss": 0.0282,
|
| 1805 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 1806 |
+
"rewards/chosen": 9.580074310302734,
|
| 1807 |
+
"rewards/margins": 42.589351654052734,
|
| 1808 |
+
"rewards/rejected": -33.00927734375,
|
| 1809 |
+
"step": 1200
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 8.013289036544851,
|
| 1813 |
+
"grad_norm": 0.09425640851259232,
|
| 1814 |
+
"learning_rate": 5.48014933308352e-07,
|
| 1815 |
+
"logits/chosen": -0.23732267320156097,
|
| 1816 |
+
"logits/rejected": -0.28549158573150635,
|
| 1817 |
+
"logps/chosen": -977.196044921875,
|
| 1818 |
+
"logps/rejected": -1548.5743408203125,
|
| 1819 |
+
"loss": 0.0337,
|
| 1820 |
+
"rewards/accuracies": 0.9490132331848145,
|
| 1821 |
+
"rewards/chosen": 9.362371444702148,
|
| 1822 |
+
"rewards/margins": 41.201656341552734,
|
| 1823 |
+
"rewards/rejected": -31.839282989501953,
|
| 1824 |
+
"step": 1210
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 8.079734219269103,
|
| 1828 |
+
"grad_norm": 0.02250111848115921,
|
| 1829 |
+
"learning_rate": 5.121980087628802e-07,
|
| 1830 |
+
"logits/chosen": -0.21824708580970764,
|
| 1831 |
+
"logits/rejected": -0.2829313278198242,
|
| 1832 |
+
"logps/chosen": -1000.27001953125,
|
| 1833 |
+
"logps/rejected": -1619.780029296875,
|
| 1834 |
+
"loss": 0.0304,
|
| 1835 |
+
"rewards/accuracies": 0.9562499523162842,
|
| 1836 |
+
"rewards/chosen": 9.380255699157715,
|
| 1837 |
+
"rewards/margins": 43.14335632324219,
|
| 1838 |
+
"rewards/rejected": -33.763099670410156,
|
| 1839 |
+
"step": 1220
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 8.146179401993356,
|
| 1843 |
+
"grad_norm": 0.0924224779009819,
|
| 1844 |
+
"learning_rate": 4.774575140626317e-07,
|
| 1845 |
+
"logits/chosen": -0.19590233266353607,
|
| 1846 |
+
"logits/rejected": -0.279987633228302,
|
| 1847 |
+
"logps/chosen": -965.008056640625,
|
| 1848 |
+
"logps/rejected": -1618.52197265625,
|
| 1849 |
+
"loss": 0.0229,
|
| 1850 |
+
"rewards/accuracies": 0.9671875238418579,
|
| 1851 |
+
"rewards/chosen": 9.053982734680176,
|
| 1852 |
+
"rewards/margins": 42.93439865112305,
|
| 1853 |
+
"rewards/rejected": -33.88041687011719,
|
| 1854 |
+
"step": 1230
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 8.212624584717608,
|
| 1858 |
+
"grad_norm": 0.030095938593149185,
|
| 1859 |
+
"learning_rate": 4.438122617983442e-07,
|
| 1860 |
+
"logits/chosen": -0.21942079067230225,
|
| 1861 |
+
"logits/rejected": -0.2864038348197937,
|
| 1862 |
+
"logps/chosen": -996.6010131835938,
|
| 1863 |
+
"logps/rejected": -1615.1669921875,
|
| 1864 |
+
"loss": 0.0283,
|
| 1865 |
+
"rewards/accuracies": 0.9593749642372131,
|
| 1866 |
+
"rewards/chosen": 9.895916938781738,
|
| 1867 |
+
"rewards/margins": 42.90495681762695,
|
| 1868 |
+
"rewards/rejected": -33.00904083251953,
|
| 1869 |
+
"step": 1240
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 8.279069767441861,
|
| 1873 |
+
"grad_norm": 0.00967055931687355,
|
| 1874 |
+
"learning_rate": 4.1128047146765936e-07,
|
| 1875 |
+
"logits/chosen": -0.21973438560962677,
|
| 1876 |
+
"logits/rejected": -0.2826644480228424,
|
| 1877 |
+
"logps/chosen": -1001.6805419921875,
|
| 1878 |
+
"logps/rejected": -1661.8216552734375,
|
| 1879 |
+
"loss": 0.0263,
|
| 1880 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1881 |
+
"rewards/chosen": 9.920458793640137,
|
| 1882 |
+
"rewards/margins": 44.114959716796875,
|
| 1883 |
+
"rewards/rejected": -34.19449996948242,
|
| 1884 |
+
"step": 1250
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 8.345514950166113,
|
| 1888 |
+
"grad_norm": 0.06534740328788757,
|
| 1889 |
+
"learning_rate": 3.798797596089351e-07,
|
| 1890 |
+
"logits/chosen": -0.21031054854393005,
|
| 1891 |
+
"logits/rejected": -0.29491880536079407,
|
| 1892 |
+
"logps/chosen": -985.116455078125,
|
| 1893 |
+
"logps/rejected": -1668.977783203125,
|
| 1894 |
+
"loss": 0.0294,
|
| 1895 |
+
"rewards/accuracies": 0.957812488079071,
|
| 1896 |
+
"rewards/chosen": 9.297616958618164,
|
| 1897 |
+
"rewards/margins": 44.02253723144531,
|
| 1898 |
+
"rewards/rejected": -34.72492218017578,
|
| 1899 |
+
"step": 1260
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 8.411960132890366,
|
| 1903 |
+
"grad_norm": 0.018059343099594116,
|
| 1904 |
+
"learning_rate": 3.4962713026158697e-07,
|
| 1905 |
+
"logits/chosen": -0.20840252935886383,
|
| 1906 |
+
"logits/rejected": -0.28752243518829346,
|
| 1907 |
+
"logps/chosen": -1003.153564453125,
|
| 1908 |
+
"logps/rejected": -1656.9368896484375,
|
| 1909 |
+
"loss": 0.0305,
|
| 1910 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1911 |
+
"rewards/chosen": 9.164114952087402,
|
| 1912 |
+
"rewards/margins": 43.66472625732422,
|
| 1913 |
+
"rewards/rejected": -34.5006103515625,
|
| 1914 |
+
"step": 1270
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 8.478405315614618,
|
| 1918 |
+
"grad_norm": 0.06759638339281082,
|
| 1919 |
+
"learning_rate": 3.2053896575809426e-07,
|
| 1920 |
+
"logits/chosen": -0.21498829126358032,
|
| 1921 |
+
"logits/rejected": -0.2865886390209198,
|
| 1922 |
+
"logps/chosen": -997.242919921875,
|
| 1923 |
+
"logps/rejected": -1629.5584716796875,
|
| 1924 |
+
"loss": 0.0336,
|
| 1925 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 1926 |
+
"rewards/chosen": 9.193347930908203,
|
| 1927 |
+
"rewards/margins": 42.87928771972656,
|
| 1928 |
+
"rewards/rejected": -33.68593978881836,
|
| 1929 |
+
"step": 1280
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 8.544850498338871,
|
| 1933 |
+
"grad_norm": 0.02950510010123253,
|
| 1934 |
+
"learning_rate": 2.9263101785268253e-07,
|
| 1935 |
+
"logits/chosen": -0.20270058512687683,
|
| 1936 |
+
"logits/rejected": -0.2855191230773926,
|
| 1937 |
+
"logps/chosen": -989.1177368164062,
|
| 1938 |
+
"logps/rejected": -1705.371337890625,
|
| 1939 |
+
"loss": 0.0317,
|
| 1940 |
+
"rewards/accuracies": 0.9546875357627869,
|
| 1941 |
+
"rewards/chosen": 9.245896339416504,
|
| 1942 |
+
"rewards/margins": 46.25841522216797,
|
| 1943 |
+
"rewards/rejected": -37.01251983642578,
|
| 1944 |
+
"step": 1290
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 8.611295681063122,
|
| 1948 |
+
"grad_norm": 0.023194456472992897,
|
| 1949 |
+
"learning_rate": 2.6591839919146963e-07,
|
| 1950 |
+
"logits/chosen": -0.22827497124671936,
|
| 1951 |
+
"logits/rejected": -0.29209384322166443,
|
| 1952 |
+
"logps/chosen": -974.6512451171875,
|
| 1953 |
+
"logps/rejected": -1619.2587890625,
|
| 1954 |
+
"loss": 0.0337,
|
| 1955 |
+
"rewards/accuracies": 0.9515624642372131,
|
| 1956 |
+
"rewards/chosen": 9.299782752990723,
|
| 1957 |
+
"rewards/margins": 42.95219039916992,
|
| 1958 |
+
"rewards/rejected": -33.65241241455078,
|
| 1959 |
+
"step": 1300
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 8.677740863787376,
|
| 1963 |
+
"grad_norm": 0.052743032574653625,
|
| 1964 |
+
"learning_rate": 2.404155751286988e-07,
|
| 1965 |
+
"logits/chosen": -0.1986398845911026,
|
| 1966 |
+
"logits/rejected": -0.28585973381996155,
|
| 1967 |
+
"logps/chosen": -958.5393676757812,
|
| 1968 |
+
"logps/rejected": -1646.5435791015625,
|
| 1969 |
+
"loss": 0.0217,
|
| 1970 |
+
"rewards/accuracies": 0.9687501192092896,
|
| 1971 |
+
"rewards/chosen": 9.183121681213379,
|
| 1972 |
+
"rewards/margins": 43.7320671081543,
|
| 1973 |
+
"rewards/rejected": -34.5489387512207,
|
| 1974 |
+
"step": 1310
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 8.744186046511627,
|
| 1978 |
+
"grad_norm": 0.0289428923279047,
|
| 1979 |
+
"learning_rate": 2.1613635589349756e-07,
|
| 1980 |
+
"logits/chosen": -0.21537990868091583,
|
| 1981 |
+
"logits/rejected": -0.28249984979629517,
|
| 1982 |
+
"logps/chosen": -1000.1814575195312,
|
| 1983 |
+
"logps/rejected": -1656.298095703125,
|
| 1984 |
+
"loss": 0.0255,
|
| 1985 |
+
"rewards/accuracies": 0.964062511920929,
|
| 1986 |
+
"rewards/chosen": 9.053632736206055,
|
| 1987 |
+
"rewards/margins": 44.6961555480957,
|
| 1988 |
+
"rewards/rejected": -35.64251708984375,
|
| 1989 |
+
"step": 1320
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 8.81063122923588,
|
| 1993 |
+
"grad_norm": 0.03176463395357132,
|
| 1994 |
+
"learning_rate": 1.9309388911139427e-07,
|
| 1995 |
+
"logits/chosen": -0.220908984541893,
|
| 1996 |
+
"logits/rejected": -0.29246407747268677,
|
| 1997 |
+
"logps/chosen": -953.7644653320312,
|
| 1998 |
+
"logps/rejected": -1591.0057373046875,
|
| 1999 |
+
"loss": 0.038,
|
| 2000 |
+
"rewards/accuracies": 0.9453125596046448,
|
| 2001 |
+
"rewards/chosen": 9.227982521057129,
|
| 2002 |
+
"rewards/margins": 43.120140075683594,
|
| 2003 |
+
"rewards/rejected": -33.89215850830078,
|
| 2004 |
+
"step": 1330
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 8.877076411960132,
|
| 2008 |
+
"grad_norm": 0.0467684343457222,
|
| 2009 |
+
"learning_rate": 1.713006526846439e-07,
|
| 2010 |
+
"logits/chosen": -0.20715700089931488,
|
| 2011 |
+
"logits/rejected": -0.279428094625473,
|
| 2012 |
+
"logps/chosen": -988.7305297851562,
|
| 2013 |
+
"logps/rejected": -1678.761474609375,
|
| 2014 |
+
"loss": 0.0197,
|
| 2015 |
+
"rewards/accuracies": 0.9718750715255737,
|
| 2016 |
+
"rewards/chosen": 9.498176574707031,
|
| 2017 |
+
"rewards/margins": 45.80161666870117,
|
| 2018 |
+
"rewards/rejected": -36.30343246459961,
|
| 2019 |
+
"step": 1340
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 8.943521594684386,
|
| 2023 |
+
"grad_norm": 0.012005633674561977,
|
| 2024 |
+
"learning_rate": 1.507684480352292e-07,
|
| 2025 |
+
"logits/chosen": -0.2161640077829361,
|
| 2026 |
+
"logits/rejected": -0.2925984859466553,
|
| 2027 |
+
"logps/chosen": -945.5839233398438,
|
| 2028 |
+
"logps/rejected": -1591.4210205078125,
|
| 2029 |
+
"loss": 0.037,
|
| 2030 |
+
"rewards/accuracies": 0.948437511920929,
|
| 2031 |
+
"rewards/chosen": 8.958191871643066,
|
| 2032 |
+
"rewards/margins": 41.801204681396484,
|
| 2033 |
+
"rewards/rejected": -32.843013763427734,
|
| 2034 |
+
"step": 1350
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 9.006644518272426,
|
| 2038 |
+
"grad_norm": 0.016238396987318993,
|
| 2039 |
+
"learning_rate": 1.31508393714177e-07,
|
| 2040 |
+
"logits/chosen": -0.22952061891555786,
|
| 2041 |
+
"logits/rejected": -0.2851778268814087,
|
| 2042 |
+
"logps/chosen": -987.5601806640625,
|
| 2043 |
+
"logps/rejected": -1585.639892578125,
|
| 2044 |
+
"loss": 0.0294,
|
| 2045 |
+
"rewards/accuracies": 0.955592155456543,
|
| 2046 |
+
"rewards/chosen": 9.82602310180664,
|
| 2047 |
+
"rewards/margins": 42.44609069824219,
|
| 2048 |
+
"rewards/rejected": -32.62006378173828,
|
| 2049 |
+
"step": 1360
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 9.073089700996677,
|
| 2053 |
+
"grad_norm": 0.022422021254897118,
|
| 2054 |
+
"learning_rate": 1.1353091938067024e-07,
|
| 2055 |
+
"logits/chosen": -0.22749130427837372,
|
| 2056 |
+
"logits/rejected": -0.29552769660949707,
|
| 2057 |
+
"logps/chosen": -969.6675415039062,
|
| 2058 |
+
"logps/rejected": -1588.7435302734375,
|
| 2059 |
+
"loss": 0.0402,
|
| 2060 |
+
"rewards/accuracies": 0.942187488079071,
|
| 2061 |
+
"rewards/chosen": 9.359356880187988,
|
| 2062 |
+
"rewards/margins": 42.008140563964844,
|
| 2063 |
+
"rewards/rejected": -32.64878845214844,
|
| 2064 |
+
"step": 1370
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 9.13953488372093,
|
| 2068 |
+
"grad_norm": 0.005254440009593964,
|
| 2069 |
+
"learning_rate": 9.684576015420277e-08,
|
| 2070 |
+
"logits/chosen": -0.2245253324508667,
|
| 2071 |
+
"logits/rejected": -0.2839411795139313,
|
| 2072 |
+
"logps/chosen": -979.505615234375,
|
| 2073 |
+
"logps/rejected": -1624.649169921875,
|
| 2074 |
+
"loss": 0.0272,
|
| 2075 |
+
"rewards/accuracies": 0.9609375596046448,
|
| 2076 |
+
"rewards/chosen": 9.78758716583252,
|
| 2077 |
+
"rewards/margins": 43.68372344970703,
|
| 2078 |
+
"rewards/rejected": -33.89613723754883,
|
| 2079 |
+
"step": 1380
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 9.205980066445182,
|
| 2083 |
+
"grad_norm": 0.02063533291220665,
|
| 2084 |
+
"learning_rate": 8.146195134284052e-08,
|
| 2085 |
+
"logits/chosen": -0.21092993021011353,
|
| 2086 |
+
"logits/rejected": -0.28807392716407776,
|
| 2087 |
+
"logps/chosen": -975.6532592773438,
|
| 2088 |
+
"logps/rejected": -1666.35107421875,
|
| 2089 |
+
"loss": 0.0261,
|
| 2090 |
+
"rewards/accuracies": 0.9625000953674316,
|
| 2091 |
+
"rewards/chosen": 9.447478294372559,
|
| 2092 |
+
"rewards/margins": 43.728919982910156,
|
| 2093 |
+
"rewards/rejected": -34.28144073486328,
|
| 2094 |
+
"step": 1390
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 9.272425249169435,
|
| 2098 |
+
"grad_norm": 0.09783101826906204,
|
| 2099 |
+
"learning_rate": 6.738782355044048e-08,
|
| 2100 |
+
"logits/chosen": -0.2123195081949234,
|
| 2101 |
+
"logits/rejected": -0.2877689301967621,
|
| 2102 |
+
"logps/chosen": -948.8867797851562,
|
| 2103 |
+
"logps/rejected": -1596.3294677734375,
|
| 2104 |
+
"loss": 0.0316,
|
| 2105 |
+
"rewards/accuracies": 0.9546874761581421,
|
| 2106 |
+
"rewards/chosen": 9.14689826965332,
|
| 2107 |
+
"rewards/margins": 42.3780517578125,
|
| 2108 |
+
"rewards/rejected": -33.23115158081055,
|
| 2109 |
+
"step": 1400
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 9.338870431893687,
|
| 2113 |
+
"grad_norm": 0.02494479902088642,
|
| 2114 |
+
"learning_rate": 5.463099816548578e-08,
|
| 2115 |
+
"logits/chosen": -0.21241770684719086,
|
| 2116 |
+
"logits/rejected": -0.2796427309513092,
|
| 2117 |
+
"logps/chosen": -1045.521484375,
|
| 2118 |
+
"logps/rejected": -1674.2296142578125,
|
| 2119 |
+
"loss": 0.0186,
|
| 2120 |
+
"rewards/accuracies": 0.973437488079071,
|
| 2121 |
+
"rewards/chosen": 9.795973777770996,
|
| 2122 |
+
"rewards/margins": 45.49605941772461,
|
| 2123 |
+
"rewards/rejected": -35.70008850097656,
|
| 2124 |
+
"step": 1410
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 9.40531561461794,
|
| 2128 |
+
"grad_norm": 0.004993575159460306,
|
| 2129 |
+
"learning_rate": 4.319838323396691e-08,
|
| 2130 |
+
"logits/chosen": -0.21156074106693268,
|
| 2131 |
+
"logits/rejected": -0.2889597713947296,
|
| 2132 |
+
"logps/chosen": -934.5477294921875,
|
| 2133 |
+
"logps/rejected": -1561.5242919921875,
|
| 2134 |
+
"loss": 0.0295,
|
| 2135 |
+
"rewards/accuracies": 0.957812488079071,
|
| 2136 |
+
"rewards/chosen": 8.935663223266602,
|
| 2137 |
+
"rewards/margins": 42.12391662597656,
|
| 2138 |
+
"rewards/rejected": -33.18825149536133,
|
| 2139 |
+
"step": 1420
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 9.471760797342192,
|
| 2143 |
+
"grad_norm": 0.003986136522144079,
|
| 2144 |
+
"learning_rate": 3.309616971855195e-08,
|
| 2145 |
+
"logits/chosen": -0.22422273457050323,
|
| 2146 |
+
"logits/rejected": -0.2854507267475128,
|
| 2147 |
+
"logps/chosen": -1002.1636352539062,
|
| 2148 |
+
"logps/rejected": -1622.54150390625,
|
| 2149 |
+
"loss": 0.0293,
|
| 2150 |
+
"rewards/accuracies": 0.9593750238418579,
|
| 2151 |
+
"rewards/chosen": 9.735203742980957,
|
| 2152 |
+
"rewards/margins": 43.484405517578125,
|
| 2153 |
+
"rewards/rejected": -33.749202728271484,
|
| 2154 |
+
"step": 1430
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 9.538205980066445,
|
| 2158 |
+
"grad_norm": 0.006039198487997055,
|
| 2159 |
+
"learning_rate": 2.4329828146074096e-08,
|
| 2160 |
+
"logits/chosen": -0.21182821691036224,
|
| 2161 |
+
"logits/rejected": -0.2917216420173645,
|
| 2162 |
+
"logps/chosen": -967.4940795898438,
|
| 2163 |
+
"logps/rejected": -1642.61474609375,
|
| 2164 |
+
"loss": 0.0349,
|
| 2165 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2166 |
+
"rewards/chosen": 9.229864120483398,
|
| 2167 |
+
"rewards/margins": 43.08539962768555,
|
| 2168 |
+
"rewards/rejected": -33.85553741455078,
|
| 2169 |
+
"step": 1440
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 9.604651162790697,
|
| 2173 |
+
"grad_norm": 0.02733450196683407,
|
| 2174 |
+
"learning_rate": 1.6904105645142443e-08,
|
| 2175 |
+
"logits/chosen": -0.2116456925868988,
|
| 2176 |
+
"logits/rejected": -0.28832632303237915,
|
| 2177 |
+
"logps/chosen": -948.6715087890625,
|
| 2178 |
+
"logps/rejected": -1624.857666015625,
|
| 2179 |
+
"loss": 0.0298,
|
| 2180 |
+
"rewards/accuracies": 0.957812488079071,
|
| 2181 |
+
"rewards/chosen": 9.185124397277832,
|
| 2182 |
+
"rewards/margins": 43.26292419433594,
|
| 2183 |
+
"rewards/rejected": -34.077796936035156,
|
| 2184 |
+
"step": 1450
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 9.67109634551495,
|
| 2188 |
+
"grad_norm": 0.01584717258810997,
|
| 2189 |
+
"learning_rate": 1.0823023375489128e-08,
|
| 2190 |
+
"logits/chosen": -0.20935006439685822,
|
| 2191 |
+
"logits/rejected": -0.2797791361808777,
|
| 2192 |
+
"logps/chosen": -1003.3470458984375,
|
| 2193 |
+
"logps/rejected": -1671.807373046875,
|
| 2194 |
+
"loss": 0.0207,
|
| 2195 |
+
"rewards/accuracies": 0.9703124761581421,
|
| 2196 |
+
"rewards/chosen": 9.804126739501953,
|
| 2197 |
+
"rewards/margins": 44.94650650024414,
|
| 2198 |
+
"rewards/rejected": -35.14237976074219,
|
| 2199 |
+
"step": 1460
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 9.737541528239202,
|
| 2203 |
+
"grad_norm": 0.06928031146526337,
|
| 2204 |
+
"learning_rate": 6.089874350439507e-09,
|
| 2205 |
+
"logits/chosen": -0.20858515799045563,
|
| 2206 |
+
"logits/rejected": -0.2875511348247528,
|
| 2207 |
+
"logps/chosen": -989.045654296875,
|
| 2208 |
+
"logps/rejected": -1670.74755859375,
|
| 2209 |
+
"loss": 0.0338,
|
| 2210 |
+
"rewards/accuracies": 0.9515625238418579,
|
| 2211 |
+
"rewards/chosen": 8.989653587341309,
|
| 2212 |
+
"rewards/margins": 45.47593307495117,
|
| 2213 |
+
"rewards/rejected": -36.48627853393555,
|
| 2214 |
+
"step": 1470
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 9.803986710963455,
|
| 2218 |
+
"grad_norm": 0.0877932533621788,
|
| 2219 |
+
"learning_rate": 2.7072216536885855e-09,
|
| 2220 |
+
"logits/chosen": -0.21547669172286987,
|
| 2221 |
+
"logits/rejected": -0.2913072109222412,
|
| 2222 |
+
"logps/chosen": -976.333740234375,
|
| 2223 |
+
"logps/rejected": -1599.1285400390625,
|
| 2224 |
+
"loss": 0.0359,
|
| 2225 |
+
"rewards/accuracies": 0.9484375715255737,
|
| 2226 |
+
"rewards/chosen": 9.221257209777832,
|
| 2227 |
+
"rewards/margins": 42.044002532958984,
|
| 2228 |
+
"rewards/rejected": -32.82274627685547,
|
| 2229 |
+
"step": 1480
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 9.870431893687707,
|
| 2233 |
+
"grad_norm": 0.000816255051176995,
|
| 2234 |
+
"learning_rate": 6.768970513457151e-10,
|
| 2235 |
+
"logits/chosen": -0.2026660144329071,
|
| 2236 |
+
"logits/rejected": -0.2809537649154663,
|
| 2237 |
+
"logps/chosen": -1004.3927001953125,
|
| 2238 |
+
"logps/rejected": -1700.61376953125,
|
| 2239 |
+
"loss": 0.0196,
|
| 2240 |
+
"rewards/accuracies": 0.971875011920929,
|
| 2241 |
+
"rewards/chosen": 9.102788925170898,
|
| 2242 |
+
"rewards/margins": 45.093666076660156,
|
| 2243 |
+
"rewards/rejected": -35.99087905883789,
|
| 2244 |
+
"step": 1490
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 9.93687707641196,
|
| 2248 |
+
"grad_norm": 0.11716156452894211,
|
| 2249 |
+
"learning_rate": 0.0,
|
| 2250 |
+
"logits/chosen": -0.21064449846744537,
|
| 2251 |
+
"logits/rejected": -0.28609463572502136,
|
| 2252 |
+
"logps/chosen": -1000.4492797851562,
|
| 2253 |
+
"logps/rejected": -1652.978515625,
|
| 2254 |
+
"loss": 0.0272,
|
| 2255 |
+
"rewards/accuracies": 0.9609375,
|
| 2256 |
+
"rewards/chosen": 9.290279388427734,
|
| 2257 |
+
"rewards/margins": 44.490787506103516,
|
| 2258 |
+
"rewards/rejected": -35.20050811767578,
|
| 2259 |
+
"step": 1500
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 9.93687707641196,
|
| 2263 |
+
"step": 1500,
|
| 2264 |
+
"total_flos": 2.9219422466201354e+19,
|
| 2265 |
+
"train_loss": 0.07426366006334623,
|
| 2266 |
+
"train_runtime": 24708.7747,
|
| 2267 |
+
"train_samples_per_second": 3.897,
|
| 2268 |
+
"train_steps_per_second": 0.061
|
| 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
|