AI RESEARCH
Minibatch Optimal Transport and Perplexity Bound Estimation in Discrete Flow Matching
arXiv CS.LG
•
ArXi:2411.00759v5 Announce Type: replace Discrete flow matching, a recent framework for modeling categorical data, has shown competitive performance with autoregressive models. However, unlike continuous flow matching, the rectification strategy cannot be applied due to the stochasticity of discrete paths, necessitating alternative methods to minimize state transitions.