UnMaskFork: Test-Time Scaling for Masked Diffusion via Deterministic Action Branching

TL;DR

UnMaskFork introduces a deterministic search-based framework using Monte Carlo Tree Search to enhance the reasoning capabilities of Masked Diffusion Language Models during inference, outperforming existing methods on complex tasks.

Contribution

The paper presents UnMaskFork, a novel test-time scaling method that employs deterministic unmasking actions and search strategies to improve MDLM performance.

Findings

Outperforms existing test-time scaling baselines on coding benchmarks.

Shows strong scalability on mathematical reasoning tasks.

Leverages iterative, non-autoregressive generation for advanced search.

Abstract

Test-time scaling strategies have effectively leveraged inference-time compute to enhance the reasoning abilities of Autoregressive Large Language Models. In this work, we demonstrate that Masked Diffusion Language Models (MDLMs) are inherently amenable to advanced search strategies, owing to their iterative and non-autoregressive generation process. To leverage this, we propose UnMaskFork (UMF), a framework that formulates the unmasking trajectory as a search tree and employs Monte Carlo Tree Search to optimize the generation path. In contrast to standard scaling methods relying on stochastic sampling, UMF explores the search space through deterministic partial unmasking actions performed by multiple MDLMs. Our empirical evaluation demonstrates that UMF consistently outperforms existing test-time scaling baselines on complex coding benchmarks, while also exhibiting strong scalability on mathematical reasoning tasks.