dLLM: Simple Diffusion Language Modeling

Lumina-DiMOO - An Open-Sourced Multi-Modal Large Diffusion Language Model

Dimple, the first Discrete Diffusion Multimodal Large Language Model

[ICCV2025] "Di[M]O: Distilling Masked Diffusion Models into One-step Generator", Yuanzhi Zhu, Xi Wang, Stéphane Lathuilière, Vicky Kalogeiton

Python package for P2 (Path Planning), a masked diffusion model sampling method for sequence generation (protein, text, etc.).

Frequently updated list of dLLM (Diffusion Large Language Models) papers, models, and other resources

To make music production easier, we introduce Amadeus , a novel MIDI generation framework. While significantly improving generation quality, we have achieved a speedup of at least 4x compared to pure autoregressive models. We will continuously update the code, models, and datasets.

🌟 Generate and understand multimodal content seamlessly with Lumina-DiMOO, an advanced large language model designed for innovative applications.