Enhancing Large Language Models in Coding Through Multi-Perspective Self-Consistency

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Huang, Baizhou, Lu, Shuai, Chen, Weizhu, Wan, Xiaojun, Duan, Nan
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910509656178688
author Huang, Baizhou
Lu, Shuai
Chen, Weizhu
Wan, Xiaojun
Duan, Nan
author_facet Huang, Baizhou
Lu, Shuai
Chen, Weizhu
Wan, Xiaojun
Duan, Nan
contents Large language models (LLMs) have exhibited remarkable ability in code generation. However, generating the correct solution in a single attempt still remains a challenge. Prior works utilize verification properties in software engineering to verify and re-rank solutions in a majority voting manner. But the assumption behind them that generated verification properties have better qualities than solutions may not always hold. In this paper, we treat them equally as different perspectives of LLMs' reasoning processes. We propose the Multi-Perspective Self-Consistency (MPSC) framework incorporating both inter- and intra-consistency across outputs from multiple perspectives. Specifically, we prompt LLMs to generate diverse outputs from three perspectives, Solution, Specification and Test case, constructing a 3-partite graph. With two measure functions of consistency, we embed both inter- and intra-consistency information into the graph. The optimal choice of solutions is then determined based on analysis in the graph. MPSC significantly boosts performance of foundation models (ChatGPT in this paper) on various benchmarks, including HumanEval (+15.91%), MBPP (+6.43%) and CodeContests (+9.37%), even surpassing GPT-4.
format Preprint
id arxiv_https___arxiv_org_abs_2309_17272
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Enhancing Large Language Models in Coding Through Multi-Perspective Self-Consistency
Huang, Baizhou
Lu, Shuai
Chen, Weizhu
Wan, Xiaojun
Duan, Nan
Computation and Language
Artificial Intelligence
Software Engineering
Large language models (LLMs) have exhibited remarkable ability in code generation. However, generating the correct solution in a single attempt still remains a challenge. Prior works utilize verification properties in software engineering to verify and re-rank solutions in a majority voting manner. But the assumption behind them that generated verification properties have better qualities than solutions may not always hold. In this paper, we treat them equally as different perspectives of LLMs' reasoning processes. We propose the Multi-Perspective Self-Consistency (MPSC) framework incorporating both inter- and intra-consistency across outputs from multiple perspectives. Specifically, we prompt LLMs to generate diverse outputs from three perspectives, Solution, Specification and Test case, constructing a 3-partite graph. With two measure functions of consistency, we embed both inter- and intra-consistency information into the graph. The optimal choice of solutions is then determined based on analysis in the graph. MPSC significantly boosts performance of foundation models (ChatGPT in this paper) on various benchmarks, including HumanEval (+15.91%), MBPP (+6.43%) and CodeContests (+9.37%), even surpassing GPT-4.
title Enhancing Large Language Models in Coding Through Multi-Perspective Self-Consistency
topic Computation and Language
Artificial Intelligence
Software Engineering
url https://arxiv.org/abs/2309.17272