Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2510.10072 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914184398110720 |
|---|---|
| author | Cai, Hua Zhao, Shuang Zhang, Liang Shen, Xuli Xu, Qing Shen, Weilin Wen, Zihao Ban, Tianke |
| author_facet | Cai, Hua Zhao, Shuang Zhang, Liang Shen, Xuli Xu, Qing Shen, Weilin Wen, Zihao Ban, Tianke |
| contents | Reasoning-focused large language models (LLMs) are rapidly evolving across various domains, yet their capabilities in handling complex legal problems remains underexplored. In this paper, we introduce Unilaw-R1, a large language model tailored for legal reasoning. With a lightweight 7-billion parameter scale, Unilaw-R1 significantly reduces deployment cost while effectively tackling three core challenges in the legal domain: insufficient legal knowledge, unreliable reasoning logic, and weak business generalization. To address these issues, we first construct Unilaw-R1-Data, a high-quality dataset containing 17K distilled and screened chain-of-thought (CoT) samples. Based on this, we adopt a two-stage training strategy combining Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL), which significantly boosts the performance on complex legal reasoning tasks and supports interpretable decision-making in legal AI applications. To assess legal reasoning ability, we also introduce Unilaw-R1-Eval, a dedicated benchmark designed to evaluate models across single- and multi-choice legal tasks. Unilaw-R1 demonstrates strong results on authoritative benchmarks, outperforming all models of similar scale and achieving performance on par with the much larger DeepSeek-R1-Distill-Qwen-32B (54.9%). Following domain-specific training, it also showed significant gains on LawBench and LexEval, exceeding Qwen-2.5-7B-Instruct (46.6%) by an average margin of 6.6%. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_10072 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Unilaw-R1: A Large Language Model for Legal Reasoning with Reinforcement Learning and Iterative Inference Cai, Hua Zhao, Shuang Zhang, Liang Shen, Xuli Xu, Qing Shen, Weilin Wen, Zihao Ban, Tianke Computation and Language Reasoning-focused large language models (LLMs) are rapidly evolving across various domains, yet their capabilities in handling complex legal problems remains underexplored. In this paper, we introduce Unilaw-R1, a large language model tailored for legal reasoning. With a lightweight 7-billion parameter scale, Unilaw-R1 significantly reduces deployment cost while effectively tackling three core challenges in the legal domain: insufficient legal knowledge, unreliable reasoning logic, and weak business generalization. To address these issues, we first construct Unilaw-R1-Data, a high-quality dataset containing 17K distilled and screened chain-of-thought (CoT) samples. Based on this, we adopt a two-stage training strategy combining Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL), which significantly boosts the performance on complex legal reasoning tasks and supports interpretable decision-making in legal AI applications. To assess legal reasoning ability, we also introduce Unilaw-R1-Eval, a dedicated benchmark designed to evaluate models across single- and multi-choice legal tasks. Unilaw-R1 demonstrates strong results on authoritative benchmarks, outperforming all models of similar scale and achieving performance on par with the much larger DeepSeek-R1-Distill-Qwen-32B (54.9%). Following domain-specific training, it also showed significant gains on LawBench and LexEval, exceeding Qwen-2.5-7B-Instruct (46.6%) by an average margin of 6.6%. |
| title | Unilaw-R1: A Large Language Model for Legal Reasoning with Reinforcement Learning and Iterative Inference |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2510.10072 |