Session 3 // Dec 15: Diffusion beats AR: Multi-modal Generation & Understanding
Diffusion LLM
Hello folks,
Discrete diffusion models offer stronger performance in multi-modal tasks, making them a compelling alternative to autoregressive models.
This Monday, John Nguyen will explain why and demonstrate how discrete diffusion models enable better multi-modal generation and understanding.
Title:
OneFlow: Concurrent Mixed-Modal and Interleaved Generation with Edit Flows
Meeting Link: click here
Time: Dec 15 (Monday) 10 am ET / 4pm CET
Paper: https://arxiv.org/abs/2510.03506
Prior knowledge:
Fundamentals of discrete diffusion
Edit Flows: Flow Matching with Edit Operations (John will give a brief introduction in the talk)
Abstract: We present OneFlow, the first non-autoregressive multimodal model that enables variable-length and concurrent mixed-modal generation. Unlike autoregressive models that enforce rigid causal ordering between text and image generation, OneFlow combines an insertion-based Edit Flow for discrete text tokens with Flow Matching for image latents. OneFlow enables concurrent text-image synthesis with hierarchical sampling that prioritizes content over grammar. Through controlled experiments across model sizes from 1B to 8B, we demonstrate that OneFlow outperforms autoregressive baselines on both generation and understanding tasks while using up to 50% fewer training FLOPs. OneFlow surpasses both autoregressive and diffusion-based approaches while unlocking new capabilities for concurrent generation, iterative refinement, and natural reasoning-like generation.
Yours truly,
Subham, Justin, Zhihan
Website, Twitter, Discord, YouTube