Fast Sampling via Discrete Non-Markov Diffusion Models with Predetermined Transition Time
This work addresses the under-explored issue of sampling acceleration for discrete diffusion models, which is incremental as it builds on existing diffusion frameworks to improve efficiency.
The paper tackles the problem of slow sampling in discrete diffusion models by proposing discrete non-Markov diffusion models (DNDM) with predetermined transition time, which enables a training-free algorithm that reduces function evaluations and accelerates generation. Experiments on natural language generation and machine translation show superior performance in speed and quality compared to existing methods.
Discrete diffusion models have emerged as powerful tools for high-quality data generation. Despite their success in discrete spaces, such as text generation tasks, the acceleration of discrete diffusion models remains under-explored. In this paper, we propose discrete non-Markov diffusion models (DNDM), which naturally induce the predetermined transition time set. This enables a training-free sampling algorithm that significantly reduces the number of function evaluations (i.e., calls to the neural network), making the sampling process much faster. Furthermore, we study the transition from finite to infinite step sampling, offering new insights into bridging the gap between discrete and continuous-time processes for discrete diffusion models. Extensive experiments on natural language generation and machine translation tasks demonstrate the superior performance of our method in terms of both generation speed and sample quality compared to existing methods for discrete diffusion models.