Model Card for simpo-math-model

This model is a fine-tuned version of rawsh/mirrorqwen2.5-0.5b-SFT. It has been trained using TRL.

Quick start

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="rawsh/simpo-math-model", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])

Training procedure

This model was trained with CPO, a method introduced in Contrastive Preference Optimization: Pushing the Boundaries of LLM Performance in Machine Translation.

Framework versions

• TRL: 0.12.0
• Transformers: 4.46.2
• Pytorch: 2.4.1
• Datasets: 3.1.0
• Tokenizers: 0.20.3

Citations

Cite CPO as:

@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:

@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}}
}