SJTU-DENG-Lab/mbd-lms

ORL: On-Policy Residual Learning for Diffusion Language Models

[ICML 2026] Set Diffusion: Interpolating Token Orderings between Autoregression and Diffusion for Fast and Flexible Decoding

Block-R1: Multi-domain RL Post-training Benchmark for dLLMs.

dLLM: Simple Diffusion Language Modeling

[ICLR 2026] Official code for TraceRL: Revolutionizing post-training for Diffusion LLMs, powering the SOTA TraDo series.

[COLM 2026] Official repository of DARE: Diffusion Large Language Models Alignment and Reinforcement Executor

Research code and optimized inference for ALoDLM: Adaptively Looped Diffusion Language Models.

Source code to accompany research paper on training multi token prediction language models using self-distillation.