@article{wang2025learning,
author = {Wang, Qibin and Zhao, Pu and Huang, Shaohan and Yang, Fangkai and Wang, Lu and Wei, Furu and Lin, Qingwei and Rajmohan, Saravan and Zhang, Dongmei},
title = {Learning to Refine: Self-Refinement of Parallel Reasoning in LLMs},
year = {2025},
month = {August},
abstract = {To further enhance the ability of Large Language Models (LLMs) to solve complex, multi-step reasoning problems, test-time scaling (TTS) methods have gained widespread attention. Existing approaches such as Best-of-N and majority voting are limited as their performance depends on the quality of candidate responses, making them unable to produce a correct solution when all candidates are incorrect. Introducing an additional model to select the best response also incurs significant deployment costs. To this end, we introduce Generative Self-Refinement (GSR), a novel parallel test-time scaling framework where a unified model first generates a set of candidate responses in parallel and then performs self-refinement to synthesize a new superior solution based on a prompt consisting of the problem and these candidates. However, LLMs struggle to perform refinement effectively when prompted directly. Therefore, we design a hybrid training pipeline by jointly optimizing for two complementary objectives, solving problems directly and refining candidate responses. Experimental results demonstrate that our method achieves state-of-the-art performance across five mathematical benchmarks. We further show that this learned self-refinement skill is a model-agnostic enhancement, robust across different model scales and generalizing to out-of-distribution reasoning tasks.},
url = {http://approjects.co.za/?big=en-us/research/publication/learning-to-refine-self-refinement-of-parallel-reasoning-in-llms/},
journal = {ArXiv},
volume = {abs/2509.00084},
}