SparseD is a novel sparse attention method for diffusion language models (DLMs), delivering near lossless acceleration in performance. It uses full attention and computes sparse patterns during early denoising steps, then reuses these patterns in later steps to restrict computation and improve…
الميزات الرئيسية لـ inv-wzq/sparsed هي: Language Diffusion Models.
تشمل البدائل مفتوحة المصدر لـ inv-wzq/sparsed: ashenweli/discrete-diffusion-models-for-language-genaration — This thesis aimstoinvestigate thepotential of discrete diffusion models in the context ofnaturallanguagegeneration. bytedance-seed/stable-diffcoder — Stable-DiffCoder: Pushing the Frontier of Code Diffusion Large Language Model. czg1225/dparallel — 🚀 dParallel: Learnable Parallel Decoding for dLLMs. dreamlm/dream-coder — Dream-Coder 7B is a diffusion LLM for code trained exclusively on open-source data across its development… dreamlm/dreamon — ](https://huggingface.co/Dream-org/DreamOn-v0-7B). aim-uofa/evotokendlm — [2026-04-07] Our Paper is accepted to ACL 2026 (main)! - [2026-01-13] Code of EvoToken-DLM Released! - [2026-01-12]…
This thesis aimstoinvestigate thepotential of discrete diffusion models in the context ofnaturallanguagegeneration.
Stable-DiffCoder: Pushing the Frontier of Code Diffusion Large Language Model
2026-04-07 Our Paper is accepted to ACL 2026 (main)! - 2026-01-13 Code of EvoToken-DLM Released! - 2026-01-12 Paper Released!
