From 14df6e6fc92d0ebb752fe7dc34aa15c1a5e5ba75 Mon Sep 17 00:00:00 2001 From: Sayak Paul Date: Sat, 14 Feb 2026 08:50:25 +0530 Subject: [PATCH] Include a note on `kernels` and `diffusers` Updated README with new information about Sage Attention support and recent releases. --- README.md | 1 + 1 file changed, 1 insertion(+) diff --git a/README.md b/README.md index 88a3fb78..a468509b 100644 --- a/README.md +++ b/README.md @@ -39,6 +39,7 @@ Paper: https://arxiv.org/abs/2505.11594 ![Local Image](./assets/4090_sageattn2++.png) +- [2025-12-20]: Sage Attention is supported via the 🤗 `kernels` library. Check it out [here](https://huggingface.co/kernels-community/sage_attention). It's also supported in the Diffusers library ([docs](https://huggingface.co/docs/diffusers/main/en/optimization/attention_backends)). - [2025-06-19]: [Sparse SageAttention1 API](https://github.com/jt-zhang/Sparse_SageAttention_API) and [Sparse SageAttention2 API](https://github.com/thu-ml/SpargeAttn) can compute attention with any block sparse pattern very fast. - [2025-05-02]: 🎉SageAttention2 and [SpargeAttn](https://github.com/thu-ml/SpargeAttn) are accepted by ICML 2025! - [2025-02-25]: 🔥 We release [SpargeAttn](https://github.com/thu-ml/SpargeAttn), a sparse attention based on SageAttention2, which could acclerate any model without training.