Model Card for Llama-3.1-Argunaut-1-8B-SPIN

Name: Llama-3.1-Argunaut-1-8B-SPIN
Availability: InStock

This model is a fine-tuned version of DebateLabKIT/Llama-3.1-Argunaut-1-8B-SFT. It has been trained using TRL and vLLM. Checkpoints are tagged.

📘 HF Blog Article

Visualize in Weights & Biases

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="DebateLabKIT/Llama-3.1-Argunaut-1-8B-SPIN", 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 Self-Play Fine-Tuning (SPIN), a method introduced in Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models.

More details about the training procedure can be found in the blog post.

Framework versions

TRL: 0.14.0
Transformers: 4.46.3
Pytorch: 2.4.1
Datasets: 3.1.0
Tokenizers: 0.20.3

Evaluation

Chat Experience


[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]
[object Object]

Metrics

	Llama-3.1-8B-Instruct	Argunaut-1-8B-SFT	Argunaut-1-8B-SPIN
⬇️ Argdown Bench
pass@1	80.8	98.9	79.5
pass@5	98.9	99.8	94.5
graph sim	8.4	65.5	34.3
isomorphic	1.3	44.0	17.1
🤗 HF Leaderboard
MMLU pro	37.6	34.5	35.9
MUSR	40.1	41.6	41.6
GPQA Diamond	32.8	30.8	31.8
GPQA Main	28.5	32.1	30.8
MATH	12.5	9.1	11.1
BBH	54.7	48.2	49.8
⛓️ COT Leaderboard
LogiQA	5.9	1.4	5.8
LogiQA2	15.5	0.8	9.2
LSAT-ar	11.7	3.0	6.5
LSAT-lr	20.8	3.9	12.9
LSAT-rc	27.5	13.8	21.2

Citations

Cite SPIN as:

bibtex

@misc{chen2024selfplayfinetuningconvertsweak,
      title={Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models}, 
      author={Zixiang Chen and Yihe Deng and Huizhuo Yuan and Kaixuan Ji and Quanquan Gu},
      year={2024},
      eprint={2401.01335},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2401.01335}, 
}

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}}
}