Open-source alternatives to Masked Diffusion Lm

Official implementation of TESS 2. TESS 2 is a state-of-the-art diffusion language model created by adapting existing pretrained autoregressive models to a diffusion paradigm. For more details, please check out our paper and model checkpoints on Hugging Face.

Official implementation of DiffusionBERT: Improving Generative Masked Language Models with Diffusion Models.

This is the official code release for the ACL Findings 2024 paper:

This is the official code release for

This is the official JAX implementation for the paper ELF: Embedded Language Flows. This code is written and tested on TPUs. A PyTorch version is available on the pytorch_elf branch.

Abstract: This paper presents the Text Encoding Diffusion Model (TEncDM), a novel approach to diffusion modeling that operates in the space of pre-trained language model encodings. In contrast to traditionally used embeddings, encodings integrate contextual information. In our approach, we also…

DiffusER: Discrete Diffusion via Edit-based Reconstruction (Reid, Hellendoorn & Neubig, 2022)

Yun-Yen Chuang · Hung-Min Hsu · Kevin Lin · Chen-Sheng Gu · Ling-Zhen Li · Ray-I Chang · Hung-yi Lee Slide Poster

Microsoft Open Source Code of Conduct

By Yuxin Chen, Chumeng Liang, Hangke Sui*, Ruihan Guo, Chaoran Cheng, Jiaxuan You, Ge Liu.

🌐 Project Page

Official Codebase for "InfoDiffusion: Information Entropy Aware Diffusion Process for Non-Autoregressive Text Generation"

Official Codebase for DiffuSeq: Sequence to Sequence Text Generation With Diffusion Models and DiffuSeq-v2: Bridging Discrete and Continuous Text Spaces for Accelerated Seq2Seq Diffusion Models.

We present a demo of SSD-LM (Semi-autoregressive Simplex-based Diffusion Language Model) in Google Colab here!

Diffusion-LM

This is the official repo for the paper SeqDiffuSeq: Text Diffusion with Encoder-Decoder Transformers

Paper Code

We introduce Text-to-text Self-conditioned Simplex Diffusion (TESS), a text diffusion model that is fully non-autoregressive, employs a new form of self-conditioning, and applies the diffusion process on the logit simplex space rather than the typical learned embedding space.

The official codebase for Empowering Diffusion Models on the Embedding Space for Text Generation.