Edit Flows: Variable Length Discrete Flow Matching with Sequence-Level Edit Operations

Marton Havasi, Brian Karrer, Itai Gat, Ricky T. Q. Chen

Keywords: Discrete Flow Matching

TL;DR: A discrete flow model with native variable-length generation capabilities using edit operations and relying only on relative token positioning.

Abstract: Autoregressive generative models naturally generate variable-length sequences, while non-autoregressive models struggle, often imposing rigid, token-wise structures. We propose Edit Flows, a non-autoregressive model that overcomes these limitations by defining a discrete flow over sequences through edit operations---insertions, deletions, and substitutions. By modeling these operations within a Continuous-time Markov Chain over the sequence space, Edit Flows enable flexible, position-relative generation that aligns more closely with the structure of sequence data. Our training method leverages an expanded state space with auxiliary variables, making the learning process efficient and tractable. Empirical results show that Edit Flows outperforms both autoregressive and mask models on image captioning and significantly outperforms the mask construction in text and code generation.

Primary Area: Probabilistic methods (e.g., variational inference, causal inference, Gaussian processes)

Submission Number: 13704