[ICLR 2026] Official repository of "Beyond Fixed: Training-Free Variable-Length Denoising for Diffusion Large Language Models"

[AAAI 2026 & ACL 2026] The official implementation of the DIFFA series for dLLM-based large audio language model

Embedding Inversion via Conditional Masked Diffusion: recover original text from embedding vectors using parallel denoising. Live demo + training pipeline + technical report.

[ICLR2026] Official Code for Loopholing Discrete Diffusion: Deterministic Bypass of the Sampling Wall

A compact, from-scratch trainer for masked diffusion and GPT-style autoregressive language models in one shared PyTorch Transformer.

[EMNLP 2026] Length-Adaptive Decoding for Masked Diffusion Machine Translation — a training-free target-length selector for masked diffusion LMs.

🛠 Reconstruct original text from text embeddings using conditional masked diffusion to reveal reversible embedding representations efficiently and accurately

A curated list of diffusion language models (dLLMs): papers, models, and code behind diffusion-based, non-autoregressive text generation, focused on 2025–2026.

Minimal masked diffusion LM in PyTorch. Pretraining + SFT + a scaling family from 50M to 350M, LLaDA recipe.

DiffusionGemma node pack for ComfyUI, built on MCP backbone for fully-agentic consumption — discrete diffusion text generation with per-step canvas snapshots, commit heatmaps, and structured trace data. Watch meaning crystallize out of noise.

Run NVIDIA's Nemotron-Labs-TwoTower-30B-A3B block-wise diffusion LM natively on Apple Silicon with MLX — AR tower + full two-tower mask-diffusion, 4/6/8-bit + bf16.

Local, type-aware diffusion code editor — on-device whole-region code edits steered by type-checker feedback inside the denoising loop (Apple Silicon / MLX)

Progressive-precision diffusion language model research with shared weights across adaptive precision routes (MLX and CUDA)

Masked-diffusion code generation for Qwen3-Coder with LoRA and verified on-policy training