Not All Tokens See Equally: Perception-Grounded Policy Optimization for Large Vision-Language Models

TL;DR

This paper introduces PGPO, a novel fine-grained policy optimization method that enhances multimodal reasoning in large vision-language models by emphasizing visually-grounded tokens, leading to significant performance improvements.

Contribution

The paper proposes PGPO, a new token-level advantage reshaping framework that improves learning signals for visually-dependent tokens in large vision-language models.

Findings

PGPO boosts model performance by 18.7% on average across benchmarks.

It reduces gradient variance and prevents training collapse.

PGPO acts as an effective regularizer for perception-grounded reasoning.

Abstract

While Reinforcement Learning from Verifiable Rewards (RLVR) has advanced reasoning in Large Vision-Language Models (LVLMs), prevailing frameworks suffer from a foundational methodological flaw: by distributing identical advantages across all generated tokens, these methods inherently dilute the learning signals essential for optimizing the critical, visually-grounded steps of multimodal reasoning. To bridge this gap, we formulate \textit{Token Visual Dependency}, quantifying the causal information gain of visual inputs via the Kullback-Leibler (KL) divergence between visual-conditioned and text-only predictive distributions. Revealing that this dependency is highly sparse and semantically pivotal, we introduce Perception-Grounded Policy Optimization (PGPO), which is a novel fine-grained credit assignment framework that dynamically reshapes advantages at the token level. Through a threshold-gated, mass-conserving mechanism, PGPO actively amplifies learning signals for visually-dependent tokens while suppressing gradient noise from linguistic priors. Extensive experiments based on the Qwen2.5-VL series across seven challenging multimodal reasoning benchmarks demonstrate that PGPO boosts models by 18.7% on average. Both theoretical and empirical analyses confirm that PGPO effectively reduces gradient variance, prevents training collapse, and acts as a potent regularizer for robust, perception-grounded multimodal reasoning. Code will be released on https://github.com/Yzk1114/PGPO.