Search papers, labs, and topics across Lattice.
This paper introduces BRAID, a novel framework that optimizes interleaved multi-modal reasoning by treating the generation of text and images as a unified Markov decision process (MDP). By enabling joint optimization through a single reinforcement learning objective, BRAID allows for coherent policy gradient propagation across both modalities, addressing the limitations of previous methods that only applied RL to text. Experimental results demonstrate that BRAID significantly outperforms existing baselines in spatial reasoning and visual perception tasks, highlighting the importance of a unified approach in multi-modal reasoning tasks.
A unified decision process for multi-modal reasoning reveals that joint optimization of text and image generation can dramatically enhance performance in complex reasoning tasks.
Unified multi-modal models (UMMs) have shown promising interleaved text-image reasoning capabilities, yet effectively optimizing such multi-turn generation via reinforcement learning (RL) remains an open challenge. Existing approaches apply RL exclusively to text steps, relegating image generation to supervised surrogates, preventing policy gradients from propagating through the full interleaved trajectory across heterogeneous modalities. This leaves the potential of RL for UMMs largely untapped. In the paper, we introduce \textbf{BRAID} (\textbf{B}ridging inte\textbf{R}le\textbf{A}ved mult\textbf{I}-modal reasoning as a unified \textbf{D}ecision process), a simple framework that casts multi-turn text-image-text reasoning as a unified Markov decision process (MDP), enabling joint optimization of textual and visual generation via a single, principled RL objective. BRAID computes a shared trajectory-level advantage and propagates it coherently into both text tokens and image denoising paths, each optimized through its modality-native policy gradient mechanism. To further address long-horizon credit assignment, BRAID employs a vision-language model (VLM) judge that scores each intermediate image on its reasoning utility, supplying dense turn-level feedback to sharpen learning at critical visual branches. Experiments on spatial reasoning and visual perception benchmarks show that BRAID consistently outperforms various baselines, confirming that a unified MDP formulation with vision-thinking guidance is essential for effective multi-modal reasoning.
UW