IDLM: Inverse-distilled Diffusion Language Models

ArXi:2602.19066v2 Announce Type: replace-cross Diffusion Language Models (DLMs) have recently achieved strong results in text generation. However, their multi-step sampling leads to slow inference, limiting practical use. To address this, we extend Inverse Distillation, a technique originally developed to accelerate continuous diffusion models, to the discrete setting. Nonetheless, this extension