[ICML 2026] d3LLM: Ultra-Fast Diffusion LLM 🚀
-
Updated
May 1, 2026 - Python
[ICML 2026] d3LLM: Ultra-Fast Diffusion LLM 🚀
Learn your codebases through self-quizzing and study. Track your knowledge coverage over time.
Minimal quickstart fork of the LLaDA repository. Watch text diffusion with a single line of code (& flag).
[ACL 2025 Oral] Official code for paper "Unifying Continuous and Discrete Text Diffusion with Non-simultaneous Diffusion Processes".
A text diffusion model based on DDPM operating in embedding space, generating coherent sequences by denoising continuous latent representations instead of discrete tokens.
Text diffusion language model: SEDD vs MDLM on a shared 77M backbone, with TensorRT FP16/FP8 inference
SwiftUI Demos for Inception Labs Diffusion Model
Text-diffusion models, quantized to run anywhere.
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.
Learning implementations of diffusion models
Matrix-style logit conditioning for DiffusionGemma's llama.cpp denoiser
Training-free control of a diffusion language model for Spanish sonnets: closed-form differentiable penalties enforce rhyme, meter and line-final repetition at sampling time, with no trained classifier. Rhyme 5.2% → 90.1% (vs 52.1% for compute-matched best-of-N); meter and fluency costs measured. Code, checkpoint, samples, raw scores.
To associate your repository with the text-diffusion topic, visit your repo's landing page and select "manage topics."