DINGO: Constrained Inference for Diffusion LLMs

TL;DR

DINGO introduces a dynamic programming-based constrained decoding method for diffusion LLMs, enabling provably distribution-preserving, user-specified constraints like regular expressions, significantly improving structured output reliability.

Contribution

The paper presents DINGO, a novel decoding algorithm that enforces constraints in diffusion LLMs while maintaining distribution fidelity, addressing a key limitation of existing models.

Findings

Achieves up to 68 percentage point improvement over unconstrained inference.

Effectively enforces user-specified regular expressions in outputs.

Maintains distribution-preserving sampling in diffusion LLMs.

Abstract

Diffusion LLMs have emerged as a promising alternative to conventional autoregressive LLMs, offering significant potential for improved runtime efficiency. However, existing diffusion models lack the ability to provably enforce user-specified formal constraints, such as regular expressions, which makes them unreliable for tasks that require structured outputs, such as fixed-schema JSON generation. Unlike autoregressive models that generate tokens sequentially, diffusion LLMs predict a block of tokens in parallel. This parallelism makes traditional constrained decoding algorithms, which are designed for sequential token prediction, ineffective at preserving the true output distribution. To address this limitation, we propose DINGO, a dynamic programming-based constrained decoding strategy that is both efficient and provably distribution-preserving. DINGO enables sampling of output strings with the highest probability under the model's predicted distribution, while strictly satisfying any user-specified regular expression. On standard symbolic math and JSON generation benchmarks, DINGO achieves up to a 68 percentage point improvement over unconstrained inference