Sherlock: Self-Correcting Reasoning in Vision-Language Models

TL;DR

Sherlock introduces a self-correcting training framework for vision-language models, enabling them to improve reasoning accuracy with minimal annotated data and outperform existing methods across multiple benchmarks.

Contribution

The paper presents Sherlock, a novel self-correction and self-improvement framework for reasoning VLMs, reducing reliance on large annotated datasets and enhancing generalization.

Findings

Achieves an average accuracy of 64.1 with direct generation and 65.4 after self-correction.

Outperforms existing models like LLaVA-CoT, Mulberry, and LlamaV-o1.

Uses less than 20% of the annotated data required by comparable methods.

Abstract

Reasoning Vision-Language Models (VLMs) have shown promising performance on complex multimodal tasks. However, they still face significant challenges: they are highly sensitive to reasoning errors, require large volumes of annotated data or accurate verifiers, and struggle to generalize beyond specific domains. To address these limitations, we explore self-correction as a strategy to enhance reasoning VLMs. We first conduct an in-depth analysis of reasoning VLMs' self-correction abilities and identify key gaps. Based on our findings, we introduce Sherlock, a self-correction and self-improvement training framework. Sherlock introduces a trajectory-level self-correction objective, a preference data construction method based on visual perturbation, and a dynamic $\beta$ for preference tuning. Once the model acquires self-correction capabilities using only 20k randomly sampled annotated data, it continues to self-improve without external supervision. Built on the Llama3.2-Vision-11B model, Sherlock achieves remarkable results across eight benchmarks, reaching an average accuracy of 64.1 with direct generation and 65.4 after self-correction. It outperforms LLaVA-CoT (63.2), Mulberry (63.9), and LlamaV-o1 (63.4) while using less than 20% of the annotated data.