analiza aiqbench
Discrete MeanFlow extends the one-step generation paradigm of continuous flow matching to discrete state spaces by modeling probability-mass transport via conditional transition kernels in continuous-time Markov chains. This theoretical framework eliminates iterative denoising or ODE integration, enabling direct sampling from a learned kernel that satisfies exact boundary conditions without auxiliary losses. The approach demonstrates high precision on synthetic sequence tasks, offering a rigorous alternative to diffusion models for discrete data generation.
Discrete MeanFlow: One-Step Generation via Conditional Transition Kernels
arXiv:2605.12805v2 Announce Type: replace-cross Abstract: MeanFlow enables one-step generation in continuous spaces by learning an average velocity over a time interval rather than the instantaneous velocity field of flow matching. However, discrete state spaces lack smooth trajectories or spatial derivatives, so the continuous formulation does not directly apply. We introduce Discrete MeanFlow, a theoretical finite-state formulation that replaces point motion with probability-mass transport. Our key object is the conditional transition kernel of a continuous-time Markov chain (CTMC), from which we define a mean discrete rate that measures average transition-probability change over a finite interval. We prove a Discrete MeanFlow identity that relates this finite-interval rate to the instantaneous CTMC generator at the endpoint, using the Kolmogorov forward equation in place of the spatial chain rule in continuous MeanFlow. We parameterize the transition kernel directly using a boundary-by-construction design that guarantees valid probability outputs and exact boundary conditions without auxiliary losses. Because the learned kernel defines a probability distribution, generation requires only a single forward pass followed by direct sampling, eliminating the need for iterative denoising, ODE integration, or multi-step refinement. We validate the identity on exact finite-state Markov chains, demonstrating that the learned kernel recovers the analytical ground truth with high precision. We also investigate the training objective and the limits of factorization on synthetic sequence tasks across varying alphabet sizes and sequence lengths.
Abstract:MeanFlow enables one-step generation in continuous spaces by learning an average velocity over a time interval rather than the instantaneous velocity field of flow matching. However, discrete state spaces lack smooth trajectories or spatial derivatives, so the continuous formulation does not directly apply. We introduce Discrete MeanFlow, a theoretical finite-state formulation that replaces point motion with probability-mass transport. Our key object is the conditional transition kernel of a continuous-time Markov chain (CTMC), from which we define a mean discrete rate that measures average transition-probability change over a finite interval. We prove a Discrete MeanFlow identity that relates this finite-interval rate to the instantaneous CTMC generator at the endpoint, using the Kolmogorov forward equation in place of the spatial chain rule in continuous MeanFlow. We parameterize the transition kernel directly using a boundary-by-construction design that guarantees valid probability outputs and exact boundary conditions without auxiliary losses. Because the learned kernel defines a probability distribution, generation requires only a single forward pass followed by direct sampling, eliminating the need for iterative denoising, ODE integration, or multi-step refinement. We validate the identity on exact finite-state Markov chains, demonstrating that the learned kernel recovers the analytical ground truth with high precision. We also investigate the training objective and the limits of factorization on synthetic sequence tasks across varying alphabet sizes and sequence lengths.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2605.12805 [cs.LG]
(or arXiv:2605.12805v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2605.12805 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 314 cuvinte
Subiecte conexe
Latent-Kernel Discrete Flow Maps for Few-Step Generation
arXiv · 31 iul. 2026
Acceleration of Diffusion Language Model through Discrete Average Generator
arXiv · 1 oct. 2026
Autoregressive One-Step Generative Modeling for Dynamical System Forecasting
arXiv · 28 iul. 2026
Discrete Wasserstein Flows for One-Step Generative Modeling
arXiv · 2 oct. 2026