Open-weight model
Sage-Attention-and-Sparge-Attention-HSWQ
by GJL ussoewwin/Sage-Attention-and-Sparge-Attention-HSWQ
Sage-Attention-and-Sparge-Attention-HSWQ is an open-weight model from GJL, released under Apache License 2.0. Its published files total 831.4 MB.
This repository contains unofficial Windows wheels for Sage-Attention 2.2＆3-1.0, Sparge-Attention-HSWQ 1.0, compiled with CUDA 13.2.
Model Card
By GJL, published under apache-2.0, revision 1d0b1df376b0.
This repository contains unofficial Windows wheels for Sage-Attention 2.2＆3-1.0, Sparge-Attention-HSWQ 1.0, compiled with CUDA 13.2. Python 3.13 (cp313) Python 3.14 (cp314) https://github.com/ussoewwin/ComfyUI-DistorchMemoryManager https://github.com/ussoewwin/SpargeAttn-hswq
Read GJL's full model card
Sage-Attention for Windows (Unofficial Build)
This repository contains unofficial Windows wheels for Sage-Attention 2.2＆3-1.0, Sparge-Attention-HSWQ 1.0, compiled with CUDA 13.2.
Supported Environments
- OS: Windows (x64)
- CUDA: 13.2
- Python:
- Python 3.13 (
cp313) - Python 3.14 (
cp314)
- Python 3.13 (
For running SA3&Sparge-Attn-HSWQ on ComfyUI
https://github.com/ussoewwin/ComfyUI-DistorchMemoryManager
About Sparge-Attention-HSWQ
https://github.com/ussoewwin/SpargeAttn-hswq
Identity and Version
- Repository
- ussoewwin/Sage-Attention-and-Sparge-Attention-HSWQ
- Publisher
- GJL
- Task
- Not stated by the source
- Modality
- Other
- Library
- Not stated by the source
- Parameters
- Not stated by the source
- Languages
- Not stated by the source
- Revision
- 1d0b1df376b0dd88ee9f5fe547a0c04c79abdfb2
- First published
- 2026-02-13
- Last updated
- 2026-10-04
Files and Weights
37 files, 831.4 MB in total.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| README.md | Documentation | 748 B | — |
| image.png | Other | 8.8 MB | 8ff33812143c |
| sageattention-2.2.0+cu130torch2.10.0-cp312-cp312-win_amd64.whl | Other | 19.1 MB | 036f786921cf |
| sageattention-2.2.0+cu130torch2.10.0-cp313-cp313-win_amd64.whl | Other | 19.1 MB | a6e00d621388 |
| sageattention-2.2.0+cu130torch2.11.0-cp313-cp313-win_amd64.whl | Other | 25.0 MB | 4dce6463bdf1 |
| sageattention-2.2.0+cu132torch2.12.0-cp312-cp312-win_amd64.whl | Other | 51.0 MB | 16f3af2aabca |
| sageattention-2.2.0+cu132torch2.12.0-cp313-cp313-win_amd64.whl | Other | 51.0 MB | e9b3e75685d9 |
| sageattention-2.2.0+cu132torch2.12.0-cp314-cp314-win_amd64.whl | Other | 51.0 MB | 239f12ee536c |
| sageattention-2.2.0+cu132torch2.12.1-cp312-cp312-win_amd64.whl | Other | 51.0 MB | 106aa81d9a39 |
| sageattention-2.2.0+cu132torch2.12.1-cp313-cp313-win_amd64.whl | Other | 51.0 MB | 8ae63e9f0eeb |
| sageattention-2.2.0+cu132torch2.12.1-cp314-cp314-win_amd64.whl | Other | 51.0 MB | 3b7af114d58e |
| sageattention-2.2.0+cu132torch2.13.0-cp312-cp312-win_amd64.whl | Other | 51.0 MB | 1b6852a3415b |
| sageattention-2.2.0+cu132torch2.13.0-cp313-cp313-win_amd64.whl | Other | 51.0 MB | 151d761266cf |
| sageattention-2.2.0+cu132torch2.13.0-cp314-cp314-win_amd64.whl | Other | 51.0 MB | 3d876b350197 |
| sageattention-2.2.0.post6+cu132torch2.13.0-cp312-cp312-win_amd64.whl | Other | 32.3 MB | e3dc2a6590bc |
| sageattention-2.2.0.post6+cu132torch2.13.0-cp313-cp313-win_amd64.whl | Other | 32.3 MB | 7086a4b4b0c1 |
| sageattention-2.2.0.post6+cu132torch2.13.0-cp314-cp314-win_amd64.whl | Other | 32.3 MB | e2f985419509 |
| sageattention-2.2.0.post6+cu132torch2.14.0-cp313-cp313-win_amd64.whl | Other | 32.3 MB | e961c911e06a |
| sageattention-2.2.0.post6+cu132torch2.14.0-cp314-cp314-linux_x86_64.whl | Other | 44.7 MB | 85f786ac9fe4 |
| sageattention-2.2.0.post6+cu132torch2.14.0-cp314-cp314-win_amd64.whl | Other | 32.3 MB | 21904bd4b5f6 |
| sageattn3-1.0.0+cu130torch2.10.0-cp312-cp312-win_amd64.whl | Other | 4.1 MB | 8728ee04ce61 |
| sageattn3-1.0.0+cu130torch2.10.0-cp313-cp313-win_amd64.whl | Other | 4.1 MB | 68360ca476b4 |
| sageattn3-1.0.0+cu130torch2.11.0-cp313-cp313-win_amd64.whl | Other | 4.1 MB | 2664025ed472 |
| sageattn3-1.0.0+cu132torch2.12.0-cp312-cp312-win_amd64.whl | Other | 3.9 MB | 44758ec3dfd8 |
| sageattn3-1.0.0+cu132torch2.12.0-cp313-cp313-win_amd64.whl | Other | 3.9 MB | 4e8b55825381 |
| sageattn3-1.0.0+cu132torch2.12.0-cp314-cp314-win_amd64.whl | Other | 4.0 MB | 41b90d5dcb65 |
| sageattn3-1.0.0+cu132torch2.12.1-cp312-cp312-win_amd64.whl | Other | 3.9 MB | f74efe76c204 |
| sageattn3-1.0.0+cu132torch2.12.1-cp313-cp313-win_amd64.whl | Other | 3.9 MB | 82100e36c59b |
| sageattn3-1.0.0+cu132torch2.12.1-cp314-cp314-win_amd64.whl | Other | 4.0 MB | f9b01ce9c5a2 |
| sageattn3-1.0.0+cu132torch2.13.0-cp312-cp312-win_amd64.whl | Other | 3.9 MB | 221e79bc26ac |
| sageattn3-1.0.0+cu132torch2.13.0-cp313-cp313-win_amd64.whl | Other | 3.9 MB | 31602a9a3f95 |
| sageattn3-1.0.0+cu132torch2.13.0-cp314-cp314-win_amd64.whl | Other | 4.0 MB | 1a408165de5e |
| sageattn3-1.0.0+cu132torch2.14.0-cp313-cp313-win_amd64.whl | Other | 3.9 MB | a5c2b8628b72 |
| sageattn3-1.0.0+cu132torch2.14.0-cp314-cp314-linux_x86_64.whl | Other | 4.2 MB | b291892fb406 |
| sageattn3-1.0.0+cu132torch2.14.0-cp314-cp314-win_amd64.whl | Other | 4.0 MB | 2200fcd87e09 |
| spas_sage_hswq_attn-1.2.0+cu132torch2.14.1cxx11abitrue-cp313-cp313-win_amd64.whl | Other | 34.4 MB | cc4cd2150c97 |
| .gitattributes | Repository | 5.3 KB | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
Released by GJL through its official repository on Hugging Face. Read the license.
Questions About Sage-Attention-and-Sparge-Attention-HSWQ
Can I use Sage-Attention-and-Sparge-Attention-HSWQ commercially?
Yes. Sage-Attention-and-Sparge-Attention-HSWQ is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.