Jaekyeom Kim
2024
Small Language Models Need Strong Verifiers to Self-Correct Reasoning
Yunxiang Zhang
|
Muhammad Khalifa
|
Lajanugen Logeswaran
|
Jaekyeom Kim
|
Moontae Lee
|
Honglak Lee
|
Lu Wang
Findings of the Association for Computational Linguistics ACL 2024
Self-correction has emerged as a promising solution to boost the reasoning performance of large language models (LLMs), where LLMs refine their solutions using self-generated critiques that pinpoint the errors. This work explores whether small (≤ 13B) language models (LMs) have the ability of self-correction on reasoning tasks with minimal inputs from stronger LMs. We propose a novel pipeline that prompts smaller LMs to collect self-correction data that supports the training of self-refinement abilities. First, we leverage correct solutions to guide the model in critiquing their incorrect responses. Second, the generated critiques, after filtering, are used for supervised fine-tuning of the self-correcting reasoner through solution refinement. Our experimental results show improved self-correction abilities of two models on five datasets spanning math and commonsense reasoning, with notable performance gains when paired with a strong GPT-4-based verifier, though limitations are identified when using a weak self-verifier for determining when to correct.
Search
Co-authors
- Yunxiang Zhang 1
- Muhammad Khalifa 1
- Lajanugen Logeswaran 1
- Moontae Lee 1
- Honglak Lee 1
- show all...
- Lu Wang 1