File size: 4,214 Bytes
e549137 d435503 e549137 d435503 e549137 d435503 e549137 d435503 e549137 d435503 e549137 d435503 e549137 20acaef e549137 20acaef e549137 20acaef e549137 20acaef e549137 20acaef bc37393 20acaef d435503 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 |
# xFormers for Windows (Torch 2.8.0.dev + CUDA 12.8 + Python 3.12)
Prebuilt `xformers` wheel for Windows 10/11 compatible with:
- **Torch**: 2.8.0.dev20250502+cu128
- **Torchvision**: 0.22.0.dev20250502+cu128
- **Torchaudio**: 2.6.0.dev20250502+cu128
- **CUDA**: 12.8
- **Python**: 3.12.x (CPython)
---
## ✅ Installation Instructions
directly install the tested version used in this build:
````
pip install https://download.pytorch.org/whl/nightly/cu128/torch-2.8.0.dev20250502%2Bcu128-cp312-cp312-win_amd64.whl
````
Install xformers Wheel
````
pip install https://huggingface.co/czmahi/xformers-windows-torch2.8-cu128-py312/resolve/main/latest-torch2.8-python3.12-xformers-comfyui-windows/xformers-0.0.31%2B8fc8ec5a.d20250503-cp312-cp312-win_amd64.whl
````
## download manully wheel from the link go the then find then downlaod and install
https://download.pytorch.org/whl/nightly/cu128
torchvision-0.22.0.dev20250502+cu128-cp312-cp312-win_amd64.whl
torchaudio-2.6.0.dev20250502+cu128-cp312-cp312-win_amd64.whl
torch-2.8.0.dev20250502+cu128-cp312-cp312-win_amd64.whl
### old version Install PyTorch Nightly (with CUDA 12.8 support)
```bash
pip install --upgrade --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128
# 🧱 xFormers for Windows: Torch 2.8.0.dev + CUDA 12.8 + Python 3.12
## ✅ Installation Guide for xFormers (Torch 2.8.0.dev + CUDA 12.8 + Python 3.12)
This prebuilt wheel is tested and built for:
- ✅ Python **3.12.x**
- ✅ PyTorch **2.8.0.dev (nightly)** with **CUDA 12.8**
- ✅ Windows 64-bit
- ✅ RTX 3000 series cards will work nividia Windows 64-bit
### 🔧 1. Install Python 3.12
Download and install from:
👉 https://www.python.org/downloads/release/python-3120/
Make sure Python and pip are available in your terminal:
```bash
python --version
pip --version
```
---
### ⚙️ 2. Install Nightly Torch, TorchVision, and TorchAudio
Use the official PyTorch nightly index for CUDA 12.8:
```bash
pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128
```
> ✅ Confirm version after install:
```bash
python -c "import torch; print(torch.__version__)"
# Should show: 2.8.0.dev + cu128
```
---
## 📦 Installation xformers:
torchaudio must install
```bash
pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128
```
```bash
pip install https://download.pytorch.org/whl/nightly/cu128/torch-2.8.0.dev20250416+cu128-cp312-cp312-win_amd64.whl --no-deps
```
download wheel for local setup: https://download.pytorch.org/whl/nightly/cu128/torch-2.8.0.dev20250416+cu128-cp312-cp312-win_amd64.whl
## 📦 Installation xformers update version :
before install update version see here for update
https://github.com/czmahi/xformers-windows-torch2.8-cu128-py312/blob/main/README_xformers_install.md
torchaudio must update
```bash
pip install --upgrade --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128
```
wheel:
Install directly via `pip` from Hugging Face:
```bash
pip install https://huggingface.co/czmahi/xformers-windows-torch2.8-cu128-py312/resolve/main/xformers-0.0.30+f2de641e.d20250418-cp312-cp312-win_amd64.whl --no-deps
```
✅ No need to build from source or downgrade PyTorch!
## ⚙️ Build Info
- Built from commit [`c5c0720`](https://github.com/facebookresearch/xformers/commit/c5c0720c0c628539affbcd5332ccc18f813ccafd)
- Compatible with:
- `torch==2.8.0.dev20250410+cu128`
- `python==3.12.0`
- `cuda==12.8`
---
## 🔖 Tags
`windows`, `torch-nightly`, `cu128`, `python312`, `xformers`, `prebuilt`, `xformers-2.8`
---
## 📄 License
This package follows the original [BSD-3-Clause License](https://github.com/facebookresearch/xformers/blob/main/LICENSE).
---
## 🧠 Credits
- Original repository: [facebookresearch/xformers](https://github.com/facebookresearch/xformers)
- Wheel hosted by: [czmahi on Hugging Face 🤗](https://huggingface.co/czmahi/xformers-windows-torch2.8-cu128-py312)
---
### ❤️ Support or Contributions
Open an issue or submit a pull request to help improve this prebuild packaging!
|