Model save
Browse files- README.md +68 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +826 -0
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: zephyr-7b-dpo-full
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- dpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for zephyr-7b-dpo-full
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="devamanyu/zephyr-7b-dpo-full", device="cuda")
|
| 24 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 25 |
+
print(output["generated_text"])
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Training procedure
|
| 29 |
+
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/devamanyu/huggingface/runs/nqluxh9k)
|
| 31 |
+
|
| 32 |
+
This model was trained with DPO, a method introduced in [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).
|
| 33 |
+
|
| 34 |
+
### Framework versions
|
| 35 |
+
|
| 36 |
+
- TRL: 0.12.0.dev0
|
| 37 |
+
- Transformers: 4.46.1
|
| 38 |
+
- Pytorch: 2.5.1
|
| 39 |
+
- Datasets: 3.1.0
|
| 40 |
+
- Tokenizers: 0.20.1
|
| 41 |
+
|
| 42 |
+
## Citations
|
| 43 |
+
|
| 44 |
+
Cite DPO as:
|
| 45 |
+
|
| 46 |
+
```bibtex
|
| 47 |
+
@inproceedings{rafailov2023direct,
|
| 48 |
+
title = {{Direct Preference Optimization: Your Language Model is Secretly a Reward Model}},
|
| 49 |
+
author = {Rafael Rafailov and Archit Sharma and Eric Mitchell and Christopher D. Manning and Stefano Ermon and Chelsea Finn},
|
| 50 |
+
year = 2023,
|
| 51 |
+
booktitle = {Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023},
|
| 52 |
+
url = {http://papers.nips.cc/paper_files/paper/2023/hash/a85b405ed65c6477a4fe8302b5e06ce7-Abstract-Conference.html},
|
| 53 |
+
editor = {Alice Oh and Tristan Naumann and Amir Globerson and Kate Saenko and Moritz Hardt and Sergey Levine},
|
| 54 |
+
}
|
| 55 |
+
```
|
| 56 |
+
|
| 57 |
+
Cite TRL as:
|
| 58 |
+
|
| 59 |
+
```bibtex
|
| 60 |
+
@misc{vonwerra2022trl,
|
| 61 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 62 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
|
| 63 |
+
year = 2020,
|
| 64 |
+
journal = {GitHub repository},
|
| 65 |
+
publisher = {GitHub},
|
| 66 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 67 |
+
}
|
| 68 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.0759161147113625,
|
| 5 |
+
"train_runtime": 3593.5482,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 17.012,
|
| 8 |
+
"train_steps_per_second": 0.133
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.46.1"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.0759161147113625,
|
| 5 |
+
"train_runtime": 3593.5482,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 17.012,
|
| 8 |
+
"train_steps_per_second": 0.133
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,826 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 478,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0020920502092050207,
|
| 13 |
+
"grad_norm": 18.792079300993226,
|
| 14 |
+
"learning_rate": 1.0416666666666666e-08,
|
| 15 |
+
"logits/chosen": -3.359375,
|
| 16 |
+
"logits/rejected": -3.34375,
|
| 17 |
+
"logps/chosen": -320.0,
|
| 18 |
+
"logps/rejected": -380.0,
|
| 19 |
+
"loss": 1.3828,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.02092050209205021,
|
| 28 |
+
"grad_norm": 18.23398541151873,
|
| 29 |
+
"learning_rate": 1.0416666666666667e-07,
|
| 30 |
+
"logits/chosen": -3.171875,
|
| 31 |
+
"logits/rejected": -3.15625,
|
| 32 |
+
"logps/chosen": -284.0,
|
| 33 |
+
"logps/rejected": -268.0,
|
| 34 |
+
"loss": 1.3841,
|
| 35 |
+
"rewards/accuracies": 0.2083333283662796,
|
| 36 |
+
"rewards/chosen": 0.00051116943359375,
|
| 37 |
+
"rewards/margins": 0.000637054443359375,
|
| 38 |
+
"rewards/rejected": -0.00012111663818359375,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.04184100418410042,
|
| 43 |
+
"grad_norm": 17.016957918520955,
|
| 44 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 45 |
+
"logits/chosen": -3.203125,
|
| 46 |
+
"logits/rejected": -3.203125,
|
| 47 |
+
"logps/chosen": -308.0,
|
| 48 |
+
"logps/rejected": -316.0,
|
| 49 |
+
"loss": 1.3834,
|
| 50 |
+
"rewards/accuracies": 0.26249998807907104,
|
| 51 |
+
"rewards/chosen": -0.000522613525390625,
|
| 52 |
+
"rewards/margins": 0.0001621246337890625,
|
| 53 |
+
"rewards/rejected": -0.000690460205078125,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.06276150627615062,
|
| 58 |
+
"grad_norm": 16.880895386487456,
|
| 59 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 60 |
+
"logits/chosen": -3.25,
|
| 61 |
+
"logits/rejected": -3.234375,
|
| 62 |
+
"logps/chosen": -324.0,
|
| 63 |
+
"logps/rejected": -282.0,
|
| 64 |
+
"loss": 1.3779,
|
| 65 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 66 |
+
"rewards/chosen": 0.0010528564453125,
|
| 67 |
+
"rewards/margins": 0.008056640625,
|
| 68 |
+
"rewards/rejected": -0.006988525390625,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.08368200836820083,
|
| 73 |
+
"grad_norm": 17.084942061574367,
|
| 74 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 75 |
+
"logits/chosen": -3.078125,
|
| 76 |
+
"logits/rejected": -3.140625,
|
| 77 |
+
"logps/chosen": -288.0,
|
| 78 |
+
"logps/rejected": -276.0,
|
| 79 |
+
"loss": 1.3604,
|
| 80 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 81 |
+
"rewards/chosen": -0.007110595703125,
|
| 82 |
+
"rewards/margins": 0.038330078125,
|
| 83 |
+
"rewards/rejected": -0.04541015625,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.10460251046025104,
|
| 88 |
+
"grad_norm": 16.90050268313775,
|
| 89 |
+
"learning_rate": 4.999733114418725e-07,
|
| 90 |
+
"logits/chosen": -3.078125,
|
| 91 |
+
"logits/rejected": -3.0,
|
| 92 |
+
"logps/chosen": -326.0,
|
| 93 |
+
"logps/rejected": -332.0,
|
| 94 |
+
"loss": 1.3289,
|
| 95 |
+
"rewards/accuracies": 0.625,
|
| 96 |
+
"rewards/chosen": -0.047607421875,
|
| 97 |
+
"rewards/margins": 0.0478515625,
|
| 98 |
+
"rewards/rejected": -0.09521484375,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.12552301255230125,
|
| 103 |
+
"grad_norm": 20.78289142989757,
|
| 104 |
+
"learning_rate": 4.990398100856366e-07,
|
| 105 |
+
"logits/chosen": -3.15625,
|
| 106 |
+
"logits/rejected": -3.09375,
|
| 107 |
+
"logps/chosen": -300.0,
|
| 108 |
+
"logps/rejected": -338.0,
|
| 109 |
+
"loss": 1.2877,
|
| 110 |
+
"rewards/accuracies": 0.699999988079071,
|
| 111 |
+
"rewards/chosen": -0.091796875,
|
| 112 |
+
"rewards/margins": 0.11279296875,
|
| 113 |
+
"rewards/rejected": -0.205078125,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.14644351464435146,
|
| 118 |
+
"grad_norm": 23.133166135835843,
|
| 119 |
+
"learning_rate": 4.967775735898179e-07,
|
| 120 |
+
"logits/chosen": -2.96875,
|
| 121 |
+
"logits/rejected": -3.03125,
|
| 122 |
+
"logps/chosen": -298.0,
|
| 123 |
+
"logps/rejected": -308.0,
|
| 124 |
+
"loss": 1.2374,
|
| 125 |
+
"rewards/accuracies": 0.75,
|
| 126 |
+
"rewards/chosen": -0.0888671875,
|
| 127 |
+
"rewards/margins": 0.1982421875,
|
| 128 |
+
"rewards/rejected": -0.287109375,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.16736401673640167,
|
| 133 |
+
"grad_norm": 38.62320509226835,
|
| 134 |
+
"learning_rate": 4.931986719649298e-07,
|
| 135 |
+
"logits/chosen": -3.109375,
|
| 136 |
+
"logits/rejected": -3.125,
|
| 137 |
+
"logps/chosen": -376.0,
|
| 138 |
+
"logps/rejected": -336.0,
|
| 139 |
+
"loss": 1.2072,
|
| 140 |
+
"rewards/accuracies": 0.6875,
|
| 141 |
+
"rewards/chosen": -0.27734375,
|
| 142 |
+
"rewards/margins": 0.26953125,
|
| 143 |
+
"rewards/rejected": -0.546875,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.18828451882845187,
|
| 148 |
+
"grad_norm": 40.482824838074976,
|
| 149 |
+
"learning_rate": 4.883222001996351e-07,
|
| 150 |
+
"logits/chosen": -3.0625,
|
| 151 |
+
"logits/rejected": -3.0625,
|
| 152 |
+
"logps/chosen": -350.0,
|
| 153 |
+
"logps/rejected": -398.0,
|
| 154 |
+
"loss": 1.1208,
|
| 155 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 156 |
+
"rewards/chosen": -0.66015625,
|
| 157 |
+
"rewards/margins": 0.5625,
|
| 158 |
+
"rewards/rejected": -1.2265625,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.20920502092050208,
|
| 163 |
+
"grad_norm": 35.10500983971475,
|
| 164 |
+
"learning_rate": 4.821741763807186e-07,
|
| 165 |
+
"logits/chosen": -2.921875,
|
| 166 |
+
"logits/rejected": -3.015625,
|
| 167 |
+
"logps/chosen": -354.0,
|
| 168 |
+
"logps/rejected": -406.0,
|
| 169 |
+
"loss": 1.1291,
|
| 170 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 171 |
+
"rewards/chosen": -0.6328125,
|
| 172 |
+
"rewards/margins": 0.6015625,
|
| 173 |
+
"rewards/rejected": -1.234375,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.20920502092050208,
|
| 178 |
+
"eval_logits/chosen": -2.96875,
|
| 179 |
+
"eval_logits/rejected": -2.953125,
|
| 180 |
+
"eval_logps/chosen": -368.0,
|
| 181 |
+
"eval_logps/rejected": -422.0,
|
| 182 |
+
"eval_loss": 0.5672109127044678,
|
| 183 |
+
"eval_rewards/accuracies": 0.71484375,
|
| 184 |
+
"eval_rewards/chosen": -0.73828125,
|
| 185 |
+
"eval_rewards/margins": 0.546875,
|
| 186 |
+
"eval_rewards/rejected": -1.28125,
|
| 187 |
+
"eval_runtime": 40.1054,
|
| 188 |
+
"eval_samples_per_second": 49.869,
|
| 189 |
+
"eval_steps_per_second": 0.798,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.2301255230125523,
|
| 194 |
+
"grad_norm": 49.983830380259825,
|
| 195 |
+
"learning_rate": 4.747874028753375e-07,
|
| 196 |
+
"logits/chosen": -2.9375,
|
| 197 |
+
"logits/rejected": -2.984375,
|
| 198 |
+
"logps/chosen": -418.0,
|
| 199 |
+
"logps/rejected": -422.0,
|
| 200 |
+
"loss": 1.1272,
|
| 201 |
+
"rewards/accuracies": 0.71875,
|
| 202 |
+
"rewards/chosen": -0.84375,
|
| 203 |
+
"rewards/margins": 0.56640625,
|
| 204 |
+
"rewards/rejected": -1.4140625,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.2510460251046025,
|
| 209 |
+
"grad_norm": 44.36112825504322,
|
| 210 |
+
"learning_rate": 4.662012913161997e-07,
|
| 211 |
+
"logits/chosen": -2.828125,
|
| 212 |
+
"logits/rejected": -2.875,
|
| 213 |
+
"logps/chosen": -380.0,
|
| 214 |
+
"logps/rejected": -418.0,
|
| 215 |
+
"loss": 1.0984,
|
| 216 |
+
"rewards/accuracies": 0.737500011920929,
|
| 217 |
+
"rewards/chosen": -0.6640625,
|
| 218 |
+
"rewards/margins": 0.6484375,
|
| 219 |
+
"rewards/rejected": -1.3125,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.2719665271966527,
|
| 224 |
+
"grad_norm": 46.661481140072915,
|
| 225 |
+
"learning_rate": 4.5646165232345103e-07,
|
| 226 |
+
"logits/chosen": -2.8125,
|
| 227 |
+
"logits/rejected": -2.890625,
|
| 228 |
+
"logps/chosen": -404.0,
|
| 229 |
+
"logps/rejected": -448.0,
|
| 230 |
+
"loss": 1.0745,
|
| 231 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 232 |
+
"rewards/chosen": -0.9765625,
|
| 233 |
+
"rewards/margins": 0.61328125,
|
| 234 |
+
"rewards/rejected": -1.5859375,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.2928870292887029,
|
| 239 |
+
"grad_norm": 42.823235443382885,
|
| 240 |
+
"learning_rate": 4.456204510851956e-07,
|
| 241 |
+
"logits/chosen": -2.84375,
|
| 242 |
+
"logits/rejected": -2.875,
|
| 243 |
+
"logps/chosen": -408.0,
|
| 244 |
+
"logps/rejected": -452.0,
|
| 245 |
+
"loss": 1.0635,
|
| 246 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 247 |
+
"rewards/chosen": -0.6875,
|
| 248 |
+
"rewards/margins": 0.6484375,
|
| 249 |
+
"rewards/rejected": -1.3359375,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.3138075313807531,
|
| 254 |
+
"grad_norm": 59.20252738056763,
|
| 255 |
+
"learning_rate": 4.337355301007335e-07,
|
| 256 |
+
"logits/chosen": -2.828125,
|
| 257 |
+
"logits/rejected": -2.78125,
|
| 258 |
+
"logps/chosen": -406.0,
|
| 259 |
+
"logps/rejected": -452.0,
|
| 260 |
+
"loss": 1.0733,
|
| 261 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 262 |
+
"rewards/chosen": -0.93359375,
|
| 263 |
+
"rewards/margins": 0.55078125,
|
| 264 |
+
"rewards/rejected": -1.484375,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.33472803347280333,
|
| 269 |
+
"grad_norm": 51.1296977483052,
|
| 270 |
+
"learning_rate": 4.2087030056579986e-07,
|
| 271 |
+
"logits/chosen": -2.8125,
|
| 272 |
+
"logits/rejected": -2.796875,
|
| 273 |
+
"logps/chosen": -376.0,
|
| 274 |
+
"logps/rejected": -440.0,
|
| 275 |
+
"loss": 1.066,
|
| 276 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 277 |
+
"rewards/chosen": -0.796875,
|
| 278 |
+
"rewards/margins": 0.83984375,
|
| 279 |
+
"rewards/rejected": -1.6328125,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.35564853556485354,
|
| 284 |
+
"grad_norm": 55.991618044626925,
|
| 285 |
+
"learning_rate": 4.070934040463998e-07,
|
| 286 |
+
"logits/chosen": -2.640625,
|
| 287 |
+
"logits/rejected": -2.609375,
|
| 288 |
+
"logps/chosen": -374.0,
|
| 289 |
+
"logps/rejected": -412.0,
|
| 290 |
+
"loss": 1.082,
|
| 291 |
+
"rewards/accuracies": 0.706250011920929,
|
| 292 |
+
"rewards/chosen": -1.0078125,
|
| 293 |
+
"rewards/margins": 0.609375,
|
| 294 |
+
"rewards/rejected": -1.609375,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.37656903765690375,
|
| 299 |
+
"grad_norm": 50.525931210242035,
|
| 300 |
+
"learning_rate": 3.9247834624635404e-07,
|
| 301 |
+
"logits/chosen": -2.625,
|
| 302 |
+
"logits/rejected": -2.765625,
|
| 303 |
+
"logps/chosen": -358.0,
|
| 304 |
+
"logps/rejected": -398.0,
|
| 305 |
+
"loss": 1.0064,
|
| 306 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 307 |
+
"rewards/chosen": -0.7890625,
|
| 308 |
+
"rewards/margins": 0.69921875,
|
| 309 |
+
"rewards/rejected": -1.484375,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.39748953974895396,
|
| 314 |
+
"grad_norm": 46.856348860317034,
|
| 315 |
+
"learning_rate": 3.7710310482256523e-07,
|
| 316 |
+
"logits/chosen": -2.671875,
|
| 317 |
+
"logits/rejected": -2.59375,
|
| 318 |
+
"logps/chosen": -390.0,
|
| 319 |
+
"logps/rejected": -448.0,
|
| 320 |
+
"loss": 1.0301,
|
| 321 |
+
"rewards/accuracies": 0.731249988079071,
|
| 322 |
+
"rewards/chosen": -0.8515625,
|
| 323 |
+
"rewards/margins": 0.65234375,
|
| 324 |
+
"rewards/rejected": -1.5078125,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.41841004184100417,
|
| 329 |
+
"grad_norm": 106.94783278420675,
|
| 330 |
+
"learning_rate": 3.610497133404795e-07,
|
| 331 |
+
"logits/chosen": -2.546875,
|
| 332 |
+
"logits/rejected": -2.640625,
|
| 333 |
+
"logps/chosen": -368.0,
|
| 334 |
+
"logps/rejected": -430.0,
|
| 335 |
+
"loss": 1.0723,
|
| 336 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 337 |
+
"rewards/chosen": -0.828125,
|
| 338 |
+
"rewards/margins": 0.73828125,
|
| 339 |
+
"rewards/rejected": -1.5703125,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.41841004184100417,
|
| 344 |
+
"eval_logits/chosen": -2.6875,
|
| 345 |
+
"eval_logits/rejected": -2.65625,
|
| 346 |
+
"eval_logps/chosen": -376.0,
|
| 347 |
+
"eval_logps/rejected": -460.0,
|
| 348 |
+
"eval_loss": 0.5170312523841858,
|
| 349 |
+
"eval_rewards/accuracies": 0.765625,
|
| 350 |
+
"eval_rewards/chosen": -0.8203125,
|
| 351 |
+
"eval_rewards/margins": 0.83203125,
|
| 352 |
+
"eval_rewards/rejected": -1.65625,
|
| 353 |
+
"eval_runtime": 39.9201,
|
| 354 |
+
"eval_samples_per_second": 50.1,
|
| 355 |
+
"eval_steps_per_second": 0.802,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.4393305439330544,
|
| 360 |
+
"grad_norm": 56.92194462169853,
|
| 361 |
+
"learning_rate": 3.4440382358952115e-07,
|
| 362 |
+
"logits/chosen": -2.671875,
|
| 363 |
+
"logits/rejected": -2.640625,
|
| 364 |
+
"logps/chosen": -402.0,
|
| 365 |
+
"logps/rejected": -438.0,
|
| 366 |
+
"loss": 1.0592,
|
| 367 |
+
"rewards/accuracies": 0.737500011920929,
|
| 368 |
+
"rewards/chosen": -0.8828125,
|
| 369 |
+
"rewards/margins": 0.734375,
|
| 370 |
+
"rewards/rejected": -1.6171875,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.4602510460251046,
|
| 375 |
+
"grad_norm": 91.8785262604961,
|
| 376 |
+
"learning_rate": 3.272542485937368e-07,
|
| 377 |
+
"logits/chosen": -2.515625,
|
| 378 |
+
"logits/rejected": -2.59375,
|
| 379 |
+
"logps/chosen": -386.0,
|
| 380 |
+
"logps/rejected": -454.0,
|
| 381 |
+
"loss": 1.036,
|
| 382 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 383 |
+
"rewards/chosen": -0.96484375,
|
| 384 |
+
"rewards/margins": 0.81640625,
|
| 385 |
+
"rewards/rejected": -1.78125,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.4811715481171548,
|
| 390 |
+
"grad_norm": 51.218599770252155,
|
| 391 |
+
"learning_rate": 3.096924887558854e-07,
|
| 392 |
+
"logits/chosen": -2.484375,
|
| 393 |
+
"logits/rejected": -2.53125,
|
| 394 |
+
"logps/chosen": -376.0,
|
| 395 |
+
"logps/rejected": -448.0,
|
| 396 |
+
"loss": 1.0678,
|
| 397 |
+
"rewards/accuracies": 0.731249988079071,
|
| 398 |
+
"rewards/chosen": -1.046875,
|
| 399 |
+
"rewards/margins": 0.79296875,
|
| 400 |
+
"rewards/rejected": -1.8359375,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.502092050209205,
|
| 405 |
+
"grad_norm": 52.33600440489297,
|
| 406 |
+
"learning_rate": 2.9181224366319943e-07,
|
| 407 |
+
"logits/chosen": -2.65625,
|
| 408 |
+
"logits/rejected": -2.546875,
|
| 409 |
+
"logps/chosen": -372.0,
|
| 410 |
+
"logps/rejected": -428.0,
|
| 411 |
+
"loss": 1.0064,
|
| 412 |
+
"rewards/accuracies": 0.762499988079071,
|
| 413 |
+
"rewards/chosen": -0.80859375,
|
| 414 |
+
"rewards/margins": 0.6953125,
|
| 415 |
+
"rewards/rejected": -1.5,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.5230125523012552,
|
| 420 |
+
"grad_norm": 51.27684803998907,
|
| 421 |
+
"learning_rate": 2.7370891215954565e-07,
|
| 422 |
+
"logits/chosen": -2.625,
|
| 423 |
+
"logits/rejected": -2.5625,
|
| 424 |
+
"logps/chosen": -424.0,
|
| 425 |
+
"logps/rejected": -478.0,
|
| 426 |
+
"loss": 1.0087,
|
| 427 |
+
"rewards/accuracies": 0.762499988079071,
|
| 428 |
+
"rewards/chosen": -1.0390625,
|
| 429 |
+
"rewards/margins": 0.8828125,
|
| 430 |
+
"rewards/rejected": -1.921875,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.5439330543933054,
|
| 435 |
+
"grad_norm": 63.264740099851814,
|
| 436 |
+
"learning_rate": 2.55479083351317e-07,
|
| 437 |
+
"logits/chosen": -2.625,
|
| 438 |
+
"logits/rejected": -2.65625,
|
| 439 |
+
"logps/chosen": -424.0,
|
| 440 |
+
"logps/rejected": -472.0,
|
| 441 |
+
"loss": 1.0101,
|
| 442 |
+
"rewards/accuracies": 0.731249988079071,
|
| 443 |
+
"rewards/chosen": -1.09375,
|
| 444 |
+
"rewards/margins": 0.875,
|
| 445 |
+
"rewards/rejected": -1.96875,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.5648535564853556,
|
| 450 |
+
"grad_norm": 43.895031684867064,
|
| 451 |
+
"learning_rate": 2.3722002126275822e-07,
|
| 452 |
+
"logits/chosen": -2.59375,
|
| 453 |
+
"logits/rejected": -2.59375,
|
| 454 |
+
"logps/chosen": -396.0,
|
| 455 |
+
"logps/rejected": -438.0,
|
| 456 |
+
"loss": 1.0423,
|
| 457 |
+
"rewards/accuracies": 0.706250011920929,
|
| 458 |
+
"rewards/chosen": -0.8984375,
|
| 459 |
+
"rewards/margins": 0.68359375,
|
| 460 |
+
"rewards/rejected": -1.5859375,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.5857740585774058,
|
| 465 |
+
"grad_norm": 53.26137168286982,
|
| 466 |
+
"learning_rate": 2.19029145890313e-07,
|
| 467 |
+
"logits/chosen": -2.546875,
|
| 468 |
+
"logits/rejected": -2.5625,
|
| 469 |
+
"logps/chosen": -374.0,
|
| 470 |
+
"logps/rejected": -426.0,
|
| 471 |
+
"loss": 1.0536,
|
| 472 |
+
"rewards/accuracies": 0.71875,
|
| 473 |
+
"rewards/chosen": -0.81640625,
|
| 474 |
+
"rewards/margins": 0.80078125,
|
| 475 |
+
"rewards/rejected": -1.6171875,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.606694560669456,
|
| 480 |
+
"grad_norm": 57.71809127266035,
|
| 481 |
+
"learning_rate": 2.0100351342479216e-07,
|
| 482 |
+
"logits/chosen": -2.640625,
|
| 483 |
+
"logits/rejected": -2.609375,
|
| 484 |
+
"logps/chosen": -386.0,
|
| 485 |
+
"logps/rejected": -450.0,
|
| 486 |
+
"loss": 1.0334,
|
| 487 |
+
"rewards/accuracies": 0.737500011920929,
|
| 488 |
+
"rewards/chosen": -1.1171875,
|
| 489 |
+
"rewards/margins": 0.75,
|
| 490 |
+
"rewards/rejected": -1.859375,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.6276150627615062,
|
| 495 |
+
"grad_norm": 59.26080657165494,
|
| 496 |
+
"learning_rate": 1.8323929841460178e-07,
|
| 497 |
+
"logits/chosen": -2.640625,
|
| 498 |
+
"logits/rejected": -2.609375,
|
| 499 |
+
"logps/chosen": -422.0,
|
| 500 |
+
"logps/rejected": -460.0,
|
| 501 |
+
"loss": 0.9917,
|
| 502 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 503 |
+
"rewards/chosen": -0.94140625,
|
| 504 |
+
"rewards/margins": 0.81640625,
|
| 505 |
+
"rewards/rejected": -1.7578125,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.6276150627615062,
|
| 510 |
+
"eval_logits/chosen": -2.625,
|
| 511 |
+
"eval_logits/rejected": -2.59375,
|
| 512 |
+
"eval_logps/chosen": -372.0,
|
| 513 |
+
"eval_logps/rejected": -464.0,
|
| 514 |
+
"eval_loss": 0.5082695484161377,
|
| 515 |
+
"eval_rewards/accuracies": 0.75390625,
|
| 516 |
+
"eval_rewards/chosen": -0.77734375,
|
| 517 |
+
"eval_rewards/margins": 0.92578125,
|
| 518 |
+
"eval_rewards/rejected": -1.703125,
|
| 519 |
+
"eval_runtime": 39.9557,
|
| 520 |
+
"eval_samples_per_second": 50.055,
|
| 521 |
+
"eval_steps_per_second": 0.801,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.6485355648535565,
|
| 526 |
+
"grad_norm": 63.68800654846809,
|
| 527 |
+
"learning_rate": 1.6583128063291573e-07,
|
| 528 |
+
"logits/chosen": -2.421875,
|
| 529 |
+
"logits/rejected": -2.53125,
|
| 530 |
+
"logps/chosen": -416.0,
|
| 531 |
+
"logps/rejected": -452.0,
|
| 532 |
+
"loss": 0.9939,
|
| 533 |
+
"rewards/accuracies": 0.75,
|
| 534 |
+
"rewards/chosen": -0.94140625,
|
| 535 |
+
"rewards/margins": 0.7734375,
|
| 536 |
+
"rewards/rejected": -1.71875,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.6694560669456067,
|
| 541 |
+
"grad_norm": 65.12912368302251,
|
| 542 |
+
"learning_rate": 1.488723393865766e-07,
|
| 543 |
+
"logits/chosen": -2.59375,
|
| 544 |
+
"logits/rejected": -2.546875,
|
| 545 |
+
"logps/chosen": -412.0,
|
| 546 |
+
"logps/rejected": -444.0,
|
| 547 |
+
"loss": 0.97,
|
| 548 |
+
"rewards/accuracies": 0.737500011920929,
|
| 549 |
+
"rewards/chosen": -0.95703125,
|
| 550 |
+
"rewards/margins": 0.80078125,
|
| 551 |
+
"rewards/rejected": -1.7578125,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.6903765690376569,
|
| 556 |
+
"grad_norm": 64.88040587462643,
|
| 557 |
+
"learning_rate": 1.3245295796480788e-07,
|
| 558 |
+
"logits/chosen": -2.625,
|
| 559 |
+
"logits/rejected": -2.578125,
|
| 560 |
+
"logps/chosen": -388.0,
|
| 561 |
+
"logps/rejected": -452.0,
|
| 562 |
+
"loss": 0.9984,
|
| 563 |
+
"rewards/accuracies": 0.668749988079071,
|
| 564 |
+
"rewards/chosen": -0.91796875,
|
| 565 |
+
"rewards/margins": 0.67578125,
|
| 566 |
+
"rewards/rejected": -1.59375,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.7112970711297071,
|
| 571 |
+
"grad_norm": 97.36761933497387,
|
| 572 |
+
"learning_rate": 1.1666074087171627e-07,
|
| 573 |
+
"logits/chosen": -2.65625,
|
| 574 |
+
"logits/rejected": -2.546875,
|
| 575 |
+
"logps/chosen": -400.0,
|
| 576 |
+
"logps/rejected": -490.0,
|
| 577 |
+
"loss": 0.9862,
|
| 578 |
+
"rewards/accuracies": 0.8125,
|
| 579 |
+
"rewards/chosen": -0.80859375,
|
| 580 |
+
"rewards/margins": 0.9921875,
|
| 581 |
+
"rewards/rejected": -1.796875,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.7322175732217573,
|
| 586 |
+
"grad_norm": 55.29816737345383,
|
| 587 |
+
"learning_rate": 1.0157994641835734e-07,
|
| 588 |
+
"logits/chosen": -2.453125,
|
| 589 |
+
"logits/rejected": -2.484375,
|
| 590 |
+
"logps/chosen": -390.0,
|
| 591 |
+
"logps/rejected": -452.0,
|
| 592 |
+
"loss": 0.9503,
|
| 593 |
+
"rewards/accuracies": 0.768750011920929,
|
| 594 |
+
"rewards/chosen": -1.0,
|
| 595 |
+
"rewards/margins": 0.87109375,
|
| 596 |
+
"rewards/rejected": -1.875,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.7531380753138075,
|
| 601 |
+
"grad_norm": 50.20493217819224,
|
| 602 |
+
"learning_rate": 8.729103716819111e-08,
|
| 603 |
+
"logits/chosen": -2.625,
|
| 604 |
+
"logits/rejected": -2.46875,
|
| 605 |
+
"logps/chosen": -438.0,
|
| 606 |
+
"logps/rejected": -486.0,
|
| 607 |
+
"loss": 1.0227,
|
| 608 |
+
"rewards/accuracies": 0.762499988079071,
|
| 609 |
+
"rewards/chosen": -1.09375,
|
| 610 |
+
"rewards/margins": 0.921875,
|
| 611 |
+
"rewards/rejected": -2.015625,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.7740585774058577,
|
| 616 |
+
"grad_norm": 52.540692008100564,
|
| 617 |
+
"learning_rate": 7.387025063449081e-08,
|
| 618 |
+
"logits/chosen": -2.46875,
|
| 619 |
+
"logits/rejected": -2.4375,
|
| 620 |
+
"logps/chosen": -420.0,
|
| 621 |
+
"logps/rejected": -448.0,
|
| 622 |
+
"loss": 1.0167,
|
| 623 |
+
"rewards/accuracies": 0.71875,
|
| 624 |
+
"rewards/chosen": -1.2109375,
|
| 625 |
+
"rewards/margins": 0.6796875,
|
| 626 |
+
"rewards/rejected": -1.890625,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.7949790794979079,
|
| 631 |
+
"grad_norm": 52.1938302087835,
|
| 632 |
+
"learning_rate": 6.138919252022435e-08,
|
| 633 |
+
"logits/chosen": -2.296875,
|
| 634 |
+
"logits/rejected": -2.34375,
|
| 635 |
+
"logps/chosen": -390.0,
|
| 636 |
+
"logps/rejected": -506.0,
|
| 637 |
+
"loss": 1.0047,
|
| 638 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 639 |
+
"rewards/chosen": -1.296875,
|
| 640 |
+
"rewards/margins": 0.94921875,
|
| 641 |
+
"rewards/rejected": -2.25,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.8158995815899581,
|
| 646 |
+
"grad_norm": 58.62340593152117,
|
| 647 |
+
"learning_rate": 4.991445467064689e-08,
|
| 648 |
+
"logits/chosen": -2.515625,
|
| 649 |
+
"logits/rejected": -2.46875,
|
| 650 |
+
"logps/chosen": -442.0,
|
| 651 |
+
"logps/rejected": -500.0,
|
| 652 |
+
"loss": 0.9798,
|
| 653 |
+
"rewards/accuracies": 0.737500011920929,
|
| 654 |
+
"rewards/chosen": -1.046875,
|
| 655 |
+
"rewards/margins": 0.85546875,
|
| 656 |
+
"rewards/rejected": -1.8984375,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.8368200836820083,
|
| 661 |
+
"grad_norm": 59.42878062470151,
|
| 662 |
+
"learning_rate": 3.9507259776993954e-08,
|
| 663 |
+
"logits/chosen": -2.515625,
|
| 664 |
+
"logits/rejected": -2.453125,
|
| 665 |
+
"logps/chosen": -404.0,
|
| 666 |
+
"logps/rejected": -490.0,
|
| 667 |
+
"loss": 1.012,
|
| 668 |
+
"rewards/accuracies": 0.75,
|
| 669 |
+
"rewards/chosen": -1.09375,
|
| 670 |
+
"rewards/margins": 0.89453125,
|
| 671 |
+
"rewards/rejected": -1.984375,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.8368200836820083,
|
| 676 |
+
"eval_logits/chosen": -2.515625,
|
| 677 |
+
"eval_logits/rejected": -2.46875,
|
| 678 |
+
"eval_logps/chosen": -386.0,
|
| 679 |
+
"eval_logps/rejected": -484.0,
|
| 680 |
+
"eval_loss": 0.4995781183242798,
|
| 681 |
+
"eval_rewards/accuracies": 0.78125,
|
| 682 |
+
"eval_rewards/chosen": -0.91015625,
|
| 683 |
+
"eval_rewards/margins": 0.984375,
|
| 684 |
+
"eval_rewards/rejected": -1.890625,
|
| 685 |
+
"eval_runtime": 39.9442,
|
| 686 |
+
"eval_samples_per_second": 50.07,
|
| 687 |
+
"eval_steps_per_second": 0.801,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.8577405857740585,
|
| 692 |
+
"grad_norm": 57.94077111648529,
|
| 693 |
+
"learning_rate": 3.022313472693447e-08,
|
| 694 |
+
"logits/chosen": -2.5625,
|
| 695 |
+
"logits/rejected": -2.53125,
|
| 696 |
+
"logps/chosen": -418.0,
|
| 697 |
+
"logps/rejected": -468.0,
|
| 698 |
+
"loss": 1.0227,
|
| 699 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 700 |
+
"rewards/chosen": -1.0,
|
| 701 |
+
"rewards/margins": 0.84375,
|
| 702 |
+
"rewards/rejected": -1.84375,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.8786610878661087,
|
| 707 |
+
"grad_norm": 54.220495002656904,
|
| 708 |
+
"learning_rate": 2.2111614344599684e-08,
|
| 709 |
+
"logits/chosen": -2.421875,
|
| 710 |
+
"logits/rejected": -2.484375,
|
| 711 |
+
"logps/chosen": -442.0,
|
| 712 |
+
"logps/rejected": -502.0,
|
| 713 |
+
"loss": 0.9755,
|
| 714 |
+
"rewards/accuracies": 0.699999988079071,
|
| 715 |
+
"rewards/chosen": -1.0234375,
|
| 716 |
+
"rewards/margins": 0.859375,
|
| 717 |
+
"rewards/rejected": -1.8828125,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.899581589958159,
|
| 722 |
+
"grad_norm": 57.28244651782191,
|
| 723 |
+
"learning_rate": 1.521597710086439e-08,
|
| 724 |
+
"logits/chosen": -2.375,
|
| 725 |
+
"logits/rejected": -2.265625,
|
| 726 |
+
"logps/chosen": -418.0,
|
| 727 |
+
"logps/rejected": -480.0,
|
| 728 |
+
"loss": 0.9483,
|
| 729 |
+
"rewards/accuracies": 0.800000011920929,
|
| 730 |
+
"rewards/chosen": -1.0703125,
|
| 731 |
+
"rewards/margins": 0.91015625,
|
| 732 |
+
"rewards/rejected": -1.9765625,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.9205020920502092,
|
| 737 |
+
"grad_norm": 63.185522298582946,
|
| 738 |
+
"learning_rate": 9.57301420397924e-09,
|
| 739 |
+
"logits/chosen": -2.484375,
|
| 740 |
+
"logits/rejected": -2.4375,
|
| 741 |
+
"logps/chosen": -408.0,
|
| 742 |
+
"logps/rejected": -460.0,
|
| 743 |
+
"loss": 0.9976,
|
| 744 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 745 |
+
"rewards/chosen": -0.9140625,
|
| 746 |
+
"rewards/margins": 0.796875,
|
| 747 |
+
"rewards/rejected": -1.7109375,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.9414225941422594,
|
| 752 |
+
"grad_norm": 72.06238722374589,
|
| 753 |
+
"learning_rate": 5.212833302556258e-09,
|
| 754 |
+
"logits/chosen": -2.484375,
|
| 755 |
+
"logits/rejected": -2.515625,
|
| 756 |
+
"logps/chosen": -420.0,
|
| 757 |
+
"logps/rejected": -510.0,
|
| 758 |
+
"loss": 0.9918,
|
| 759 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 760 |
+
"rewards/chosen": -1.0,
|
| 761 |
+
"rewards/margins": 0.8125,
|
| 762 |
+
"rewards/rejected": -1.8125,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.9623430962343096,
|
| 767 |
+
"grad_norm": 63.02015092631578,
|
| 768 |
+
"learning_rate": 2.158697848236607e-09,
|
| 769 |
+
"logits/chosen": -2.484375,
|
| 770 |
+
"logits/rejected": -2.453125,
|
| 771 |
+
"logps/chosen": -396.0,
|
| 772 |
+
"logps/rejected": -440.0,
|
| 773 |
+
"loss": 0.9981,
|
| 774 |
+
"rewards/accuracies": 0.731249988079071,
|
| 775 |
+
"rewards/chosen": -0.95703125,
|
| 776 |
+
"rewards/margins": 0.82421875,
|
| 777 |
+
"rewards/rejected": -1.78125,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.9832635983263598,
|
| 782 |
+
"grad_norm": 58.50751073465125,
|
| 783 |
+
"learning_rate": 4.269029751107489e-10,
|
| 784 |
+
"logits/chosen": -2.4375,
|
| 785 |
+
"logits/rejected": -2.4375,
|
| 786 |
+
"logps/chosen": -404.0,
|
| 787 |
+
"logps/rejected": -498.0,
|
| 788 |
+
"loss": 0.9687,
|
| 789 |
+
"rewards/accuracies": 0.762499988079071,
|
| 790 |
+
"rewards/chosen": -0.93359375,
|
| 791 |
+
"rewards/margins": 0.8984375,
|
| 792 |
+
"rewards/rejected": -1.8359375,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 1.0,
|
| 797 |
+
"step": 478,
|
| 798 |
+
"total_flos": 0.0,
|
| 799 |
+
"train_loss": 1.0759161147113625,
|
| 800 |
+
"train_runtime": 3593.5482,
|
| 801 |
+
"train_samples_per_second": 17.012,
|
| 802 |
+
"train_steps_per_second": 0.133
|
| 803 |
+
}
|
| 804 |
+
],
|
| 805 |
+
"logging_steps": 10,
|
| 806 |
+
"max_steps": 478,
|
| 807 |
+
"num_input_tokens_seen": 0,
|
| 808 |
+
"num_train_epochs": 1,
|
| 809 |
+
"save_steps": 100,
|
| 810 |
+
"stateful_callbacks": {
|
| 811 |
+
"TrainerControl": {
|
| 812 |
+
"args": {
|
| 813 |
+
"should_epoch_stop": false,
|
| 814 |
+
"should_evaluate": false,
|
| 815 |
+
"should_log": false,
|
| 816 |
+
"should_save": true,
|
| 817 |
+
"should_training_stop": true
|
| 818 |
+
},
|
| 819 |
+
"attributes": {}
|
| 820 |
+
}
|
| 821 |
+
},
|
| 822 |
+
"total_flos": 0.0,
|
| 823 |
+
"train_batch_size": 8,
|
| 824 |
+
"trial_name": null,
|
| 825 |
+
"trial_params": null
|
| 826 |
+
}
|