Improving Text Style Transfer using Masked Diffusion Language Models with Inference-time Scaling

2025-08-14

AI Summary

This work addresses the limitations of autoregressive language models in text style transfer—namely, constrained generation quality and low inference efficiency. We propose a novel non-autoregressive generation framework built upon Masked Diffusion Models (MDMs) and inference-time scaling. Its core innovation is a soft-value validator grounded in pretrained text embeddings, which replaces conventional classifier-free guidance to enable fine-grained, differentiable scoring of style consistency and gradient-based optimization. This validator facilitates precise alignment between generated outputs and target styles while preserving semantic fidelity. Empirical evaluation across multiple standard style transfer benchmarks demonstrates substantial improvements in both style accuracy and semantic preservation, outperforming state-of-the-art autoregressive and non-autoregressive baselines. The results validate MDMs as an effective and superior paradigm for efficient, controllable text generation.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Masked diffusion language models (MDMs) have recently gained traction as a viable generative framework for natural language. This can be attributed to its scalability and ease of training compared to other diffusion model paradigms for discrete data, establishing itself as the state-of-the-art non-autoregressive generator for discrete data. Diffusion models, in general, have shown excellent ability to improve the generation quality by leveraging inference-time scaling either by increasing the number of denoising steps or by using external verifiers on top of the outputs of each step to guide the generation. In this work, we propose a verifier-based inference-time scaling method that aids in finding a better candidate generation during the denoising process of the MDM. Our experiments demonstrate the application of MDMs for standard text-style transfer tasks and establish MDMs as a better alternative to autoregressive language models. Additionally, we show that a simple soft-value-based verifier setup for MDMs using off-the-shelf pre-trained embedding models leads to significant gains in generation quality even when used on top of typical classifier-free guidance setups in the existing literature.

Problem

• Enhancing text style transfer via masked diffusion models
• Improving generation quality with inference-time scaling
• Replacing autoregressive models with verifier-guided MDMs

Innovation

• Masked diffusion language models for text transfer
• Inference-time scaling with verifier guidance
• Soft-value verifier using pre-trained embeddings