Diffusion language models speed token generation with overlapping blocks

Blurring the Block Boundaries: Overlapping Block Denoising for Efficient Diffusion Language Models

Computation and LanguageArtificial Intelligence

Summary

Diffusion language models generate text faster by refining words in chunks called blocks. Usually, each block waits for the slowest word to finish before moving on, which can be slow. The paper presents a way to blur the edges between these blocks by working on two blocks at the same time in one step. This method speeds up text generation by about 12-15% without hurting quality, and it works without any extra training or setup.

What this means in practice

  • For mobile app developers: Create apps that generate text responses more quickly by integrating the overlapping block refinement method in diffusion language models.
  • For chatbot engineers: Improve response speed of chatbots without loss in reply quality by applying simultaneous block token refinement during text generation.

Authors

Amir Mann, Jonathan Fhima, Shaked Brody, Or Avitan, Asif Zamir, Shahar Tsiper, Niv Nayman

Abstract

Diffusion Language Models (DLMs) have emerged as effective alternatives to autoregressive models for latency-critical tasks, enabling faster token generation. Recent advances employ block diffusion with self-editing capabilities, where each block refines tokens iteratively before advancing to the next one. Yet each block is only as fast as its slowest token, and the next block cannot start before it. We suggest Blurring the Block Boundaries (B3), a novel method with which two blocks are refined in parallel within a shared forward pass. B3 calibrates itself on the fly, opening the next block only once the model is ready to start generating it. It is a training-free and hyperparameter-free acceleration method which seamlessly integrates within any self-editing block diffusion model. Evaluated on state-of-the-art DLMs across 10 diverse datasets spanning varied tasks and difficulty levels, B3 results in on par generation quality and 11.5-14.5% speedup compared to the base model, pushing the Pareto frontier for latency-quality trade-offs in DLM inference.