Spaces:
Running
Running
File size: 2,821 Bytes
dcf0937 |
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 |
ο»Ώ# π§ Installation Guide - Fixing Dependency Issues
## Problem
The error you encountered is due to `flash-attn` requiring the `packaging` module during compilation, and it's a notoriously difficult package to install on some systems.
## Solution
### Option 1: Use the Safe Installation Script (Recommended)
**For Windows:**
```powershell
# Run the safe installation script
.\install_dependencies.ps1
```
**For Linux/Mac:**
```bash
# Run the safe installation script
python install_dependencies.py
```
### Option 2: Manual Installation Steps
1. **Upgrade pip and build tools:**
```bash
pip install --upgrade pip setuptools wheel packaging
```
2. **Install PyTorch first:**
```bash
# For CUDA support
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124
# Or CPU-only version
pip install torch torchvision torchaudio
```
3. **Install main requirements (flash-attn excluded):**
```bash
pip install -r requirements.txt
```
4. **Optional: Install performance packages manually:**
```bash
# xformers (usually works)
pip install xformers
# flash-attn (may fail - it's optional)
pip install flash-attn --no-build-isolation
```
### Option 3: Skip Problematic Dependencies
The app will work perfectly fine without `flash-attn` and `xformers`. These are performance optimizations, not requirements.
## What Changed
β
**Fixed requirements.txt:**
- Added essential build dependencies (`setuptools`, `wheel`, `packaging`)
- Commented out problematic packages (`flash-attn`, `xformers`)
- Made numpy version compatible
- Added proper PyTorch installation notes
β
**Created safe installation scripts:**
- `install_dependencies.py` - Cross-platform Python script
- `install_dependencies.ps1` - Windows PowerShell script
- Both handle errors gracefully and skip optional packages
## Verification
After installation, verify everything works:
```bash
python -c "import torch, transformers, gradio, fastapi; print('β
Core dependencies installed!')"
```
## Next Steps
Once dependencies are installed:
1. **Download OmniAvatar models:**
```bash
python setup_omniavatar.py
```
2. **Start the application:**
```bash
python app.py
```
## Troubleshooting
**If you still get errors:**
1. **Use a virtual environment:**
```bash
python -m venv omniavatar_env
source omniavatar_env/bin/activate # Linux/Mac
# or
omniavatar_env\Scripts\activate # Windows
```
2. **Try without optional packages:**
The app will work fine with just the core dependencies. Performance optimizations like `flash-attn` are nice-to-have, not essential.
3. **Check Python version:**
Ensure you're using Python 3.8 or later:
```bash
python --version
```
The dependency issues have been resolved and the OmniAvatar integration will work with or without the optional performance packages! π
|