Relation-R1

🛠️ Installation

To install using pip:

pip install ms-swift -U

To install from source:

# pip install git+https://github.com/modelscope/ms-swift.git

git clone git@github.com:HKUST-LongGroup/Relation-R1.git #3.3 version
pip install -e .

bash requirements/install_all.sh

Running Environment:

	Range	Recommended	Notes
python	>=3.9	3.10
cuda		cuda12	No need to install if using CPU, NPU, MPS
torch	>=2.0
transformers	>=4.33	4.49
modelscope	>=1.19
peft	>=0.11,<0.15
trl	>=0.13,<0.17	0.15	RLHF
deepspeed	>=0.14	0.14.5	Training
vllm	>=0.5.1	0.7.3	Inference/Deployment/Evaluation
lmdeploy	lmdeploy>=0.5	0.7.1	Inference/Deployment/Evaluation
evalscope	>=0.11		Evaluation

For more optional dependencies, you can refer to here.

🚀 Quick Start

Command Train

Execute the training code for all stages according to the order of the shell scripts in the useful/train directory. For example:

bash useful_sh/train/1.fix_sft.sh

Command SGG/GSR Infer

bash useful_sh/infer/sgg_infer.sh

Command SGG/GSR Eval

bash useful_sh/eval/sgg_eval.sh

🗓️ TODO

• Release initial code
• Release training/test data
• Release Relation-R1 checkpoint

Acknowledgement

We thank:

• ms-swift
• ASv2 for their awesome open-source codes and models.

🖊️ BibTeX

If you find this project useful in your research, please consider cite:

@inproceedings{Lin2025relr1,
title={Relation-R1: Progressively Cognitive Chain-of-Thought Guided Reinforcement Learning for Unified Relation Comprehension},
booktitle={AAAI Conference on Artificial Intelligence},
author={Lin Li and Wei Chen and Jiahui Li and Kwang-Ting Cheng and Long Chen},
date={2026},
}