|
|
--- |
|
|
base_model: Aratako/gemma-2-2b-axolotl-sft-v1.0-merged |
|
|
library_name: transformers |
|
|
model_name: gemma-2-2b-axolotl-simpo-v1.0 |
|
|
tags: |
|
|
- generated_from_trainer |
|
|
- axolotl |
|
|
- trl |
|
|
- cpo |
|
|
licence: license |
|
|
--- |
|
|
|
|
|
本モデルはaxolotlの使い方の解説記事のデモで作成されたモデルです。モデルとしては特に特に利用価値のないものになっているのでご注意ください。 |
|
|
|
|
|
[記事リンクはこちら](https://zenn.dev/aratako_lm/articles/b58ac364f9c9cd) |
|
|
|
|
|
以下、自動生成されたREADMEです。 |
|
|
|
|
|
# Model Card for gemma-2-2b-axolotl-simpo-v1.0 |
|
|
|
|
|
This model is a fine-tuned version of [Aratako/gemma-2-2b-axolotl-sft-v1.0-merged](https://huggingface.co/Aratako/gemma-2-2b-axolotl-sft-v1.0-merged). |
|
|
It has been trained using [TRL](https://github.com/huggingface/trl). |
|
|
|
|
|
## Quick start |
|
|
|
|
|
```python |
|
|
from transformers import pipeline |
|
|
|
|
|
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?" |
|
|
generator = pipeline("text-generation", model="Aratako/gemma-2-2b-axolotl-simpo-v1.0", device="cuda") |
|
|
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0] |
|
|
print(output["generated_text"]) |
|
|
``` |
|
|
|
|
|
## Training procedure |
|
|
|
|
|
[<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/aratako-lm/axolotl/runs/mz5mncpr) |
|
|
|
|
|
This model was trained with CPO, a method introduced in [Contrastive Preference Optimization: Pushing the Boundaries of LLM Performance in Machine Translation](https://huggingface.co/papers/2401.08417). |
|
|
|
|
|
### Framework versions |
|
|
|
|
|
- TRL: 0.12.1 |
|
|
- Transformers: 4.46.3 |
|
|
- Pytorch: 2.5.1 |
|
|
- Datasets: 3.1.0 |
|
|
- Tokenizers: 0.20.3 |
|
|
|
|
|
## Citations |
|
|
|
|
|
Cite CPO as: |
|
|
|
|
|
```bibtex |
|
|
@inproceedings{xu2024contrastive, |
|
|
title = {{Contrastive Preference Optimization: Pushing the Boundaries of LLM Performance in Machine Translation}}, |
|
|
author = {Haoran Xu and Amr Sharaf and Yunmo Chen and Weiting Tan and Lingfeng Shen and Benjamin Van Durme and Kenton Murray and Young Jin Kim}, |
|
|
year = 2024, |
|
|
booktitle = {Forty-first International Conference on Machine Learning, {ICML} 2024, Vienna, Austria, July 21-27, 2024}, |
|
|
publisher = {OpenReview.net}, |
|
|
url = {https://openreview.net/forum?id=51iwkioZpn} |
|
|
} |
|
|
``` |
|
|
|
|
|
Cite TRL as: |
|
|
|
|
|
```bibtex |
|
|
@misc{vonwerra2022trl, |
|
|
title = {{TRL: Transformer Reinforcement Learning}}, |
|
|
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}, |
|
|
year = 2020, |
|
|
journal = {GitHub repository}, |
|
|
publisher = {GitHub}, |
|
|
howpublished = {\url{https://github.com/huggingface/trl}} |
|
|
} |
|
|
``` |
