upload torch-2.10.0a0+gite67e3d9-cp310-cp310-win_amd64.whl

---
license: other
license_name: pytorch-bsd-nvidia-cuda-eula
license_link: https://github.com/pytorch/pytorch/blob/main/LICENSE
tags:
- pytorch
- rtx-5080
- cuda-13
- windows-11
- blackwell
- sm-120
---

# PyTorch 2.10 for NVIDIA RTX 5080 - Windows 11

**Unofficial build with CUDA 13.0 support for Blackwell architecture (SM 120)**

## The Problem

- 📅 RTX 5080 launched: January 2025
- 📅 Official PyTorch support: Expected 2026
- ⏳ **You've been waiting 9+ months**

## The Solution

This pre-built package works **TODAY** with your RTX 5080.

## Download & Install

### Requirements

- NVIDIA GeForce RTX 5080
- Windows 11 (64-bit)
- Python 3.10 or 3.11
- Latest NVIDIA drivers (560+)
- ~15GB free disk space

### Installation Steps

1. **Download:**
- Click "Files and versions" above
- Download `pytorch-rtx5080-windows11.zip` (6GB)

2. **Extract:**
```powershell
# Extract to a folder with plenty of space
```

3. **Install:**
```powershell
# Navigate to extracted folder
cd pytorch-rtx5080-windows11

# Create virtual environment
python -m venv .venv
.\.venv\Scripts\Activate.ps1

# Run installer
.\install.ps1
```

4. **Verify:**
```powershell
python -c "import torch; print('CUDA:', torch.cuda.is_available())"
# Should print: CUDA: True
```

## What's Included

- PyTorch 2.10.0a0 (commit: e67e3d9)
- CUDA 13.0 runtime libraries (cuBLAS, cuDNN, etc.)
- All required DLL dependencies
- Installation script

## Package Size

**6GB compressed, ~15GB extracted**

Why so large? Includes all CUDA runtime libraries so you don't need to install CUDA Toolkit separately.

## Tested With

- ✅ Stable Diffusion WebUI
- ✅ ComfyUI
- ✅ PyTorch basic operations
- ✅ Model training
- ✅ Inference workloads

## Build Information

- **Source:** PyTorch main branch
- **CUDA:** 13.0.48
- **Python:** 3.10.11
- **Compiler:** MSVC 14.44
- **Compute Capability:** SM 120 (Blackwell)
- **Build Date:** October 2025

## Troubleshooting

### "CUDA not available"
- Update NVIDIA drivers to 560+
- Verify RTX 5080 is detected in Device Manager
- Run as administrator

### Import errors
- Ensure Python 3.10 or 3.11
- Use a fresh virtual environment
- Run `install.ps1` again

## When Will Official Support Arrive?

Official PyTorch support for RTX 5080 is expected in 2026. This package will be maintained until then.

## Disclaimer

⚠️ **Unofficial build** - not affiliated with PyTorch or NVIDIA

- Use at your own risk
- Not recommended for production
- No warranty or support guarantees
- For development and experimentation

## License

PyTorch is BSD-licensed. CUDA libraries are subject to NVIDIA EULA.

## Community & Support

- GitHub: [Your GitHub link when you make it]
- Issues: Report problems via GitHub Issues
- Discussions: Share results and help others

---

*Built by the community, for the community. Don't wait until 2026.*
```

## Step 5: Get Your Download Link

After upload completes, your download link will be:
```
https://huggingface.co/your-username/pytorch-rtx5080-windows11/resolve/main/pytorch-rtx5080-windows11.zip
```

**Or click on the file in the "Files" tab and copy the download link.**

## Step 6: Set Up Additional Details (Optional but Good)

1. Click **"Settings"** tab
2. Add tags: `pytorch`, `cuda`, `rtx-5080`, `windows`
3. Save changes

## Visual Guide - What to Click:
```
HuggingFace.co
├── Sign Up → Create account
├── [Your Profile Icon] → New Model
│ └── Fill form:
│ ├── Name: pytorch-rtx5080-windows11
│ ├── License: other
│ └── Public: ✓
├── Files and versions → Add file → Upload files
│ └── Drag pytorch-rtx5080-windows11.zip
└── Edit model card → Paste README above

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+torch-2.10.0a0+gite67e3d9-cp310-cp310-win_amd64.whl filter=lfs diff=lfs merge=lfs -text

torch-2.10.0a0+gite67e3d9-cp310-cp310-win_amd64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6202cfa3f4dac89e87bd21b754b3778288849428576e1bfd1dc11de4cfee421d
+size 192486657