Vision-Language Models Can Self-Improve Reasoning via Reflection

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cheng, Kanzhi, Li, Yantao, Xu, Fangzhi, Zhang, Jianbing, Zhou, Hao, Liu, Yang
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910683328675840
author Cheng, Kanzhi
Li, Yantao
Xu, Fangzhi
Zhang, Jianbing
Zhou, Hao
Liu, Yang
author_facet Cheng, Kanzhi
Li, Yantao
Xu, Fangzhi
Zhang, Jianbing
Zhou, Hao
Liu, Yang
contents Chain-of-thought (CoT) has proven to improve the reasoning capability of large language models (LLMs). However, due to the complexity of multimodal scenarios and the difficulty in collecting high-quality CoT data, CoT reasoning in multimodal LLMs has been largely overlooked. To this end, we propose a simple yet effective self-training framework, R3V, which iteratively enhances the model's Vision-language Reasoning by Reflecting on CoT Rationales. Our framework consists of two interleaved parts: (1) iteratively bootstrapping positive and negative solutions for reasoning datasets, and (2) reflection on rationale for learning from mistakes. Specifically, we introduce the self-refine and self-select losses, enabling the model to refine flawed rationale and derive the correct answer by comparing rationale candidates. Experiments on a wide range of vision-language tasks show that R3V consistently improves multimodal LLM reasoning, achieving a relative improvement of 23 to 60 percent over GPT-distilled baselines. Additionally, our approach supports self-reflection on generated solutions, further boosting performance through test-time computation.
format Preprint
id arxiv_https___arxiv_org_abs_2411_00855
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Vision-Language Models Can Self-Improve Reasoning via Reflection
Cheng, Kanzhi
Li, Yantao
Xu, Fangzhi
Zhang, Jianbing
Zhou, Hao
Liu, Yang
Machine Learning
Artificial Intelligence
Computation and Language
Computer Vision and Pattern Recognition
Chain-of-thought (CoT) has proven to improve the reasoning capability of large language models (LLMs). However, due to the complexity of multimodal scenarios and the difficulty in collecting high-quality CoT data, CoT reasoning in multimodal LLMs has been largely overlooked. To this end, we propose a simple yet effective self-training framework, R3V, which iteratively enhances the model's Vision-language Reasoning by Reflecting on CoT Rationales. Our framework consists of two interleaved parts: (1) iteratively bootstrapping positive and negative solutions for reasoning datasets, and (2) reflection on rationale for learning from mistakes. Specifically, we introduce the self-refine and self-select losses, enabling the model to refine flawed rationale and derive the correct answer by comparing rationale candidates. Experiments on a wide range of vision-language tasks show that R3V consistently improves multimodal LLM reasoning, achieving a relative improvement of 23 to 60 percent over GPT-distilled baselines. Additionally, our approach supports self-reflection on generated solutions, further boosting performance through test-time computation.
title Vision-Language Models Can Self-Improve Reasoning via Reflection
topic Machine Learning
Artificial Intelligence
Computation and Language
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2411.00855