Seed-X Translation GUI

Professional Python GUI application for using the Seed-X-PPO-7B translation model.

Features

• Support for translations between 28 languages
• Modern GUI interface with PyQt6
• Multi-threaded model loading and translation (non-blocking UI)
• Chain-of-Thought (CoT) mode with detailed explanations
• Adjustable generation parameters (temperature, top-p, top-k, etc.)
• Translation history with export/import functionality
• Dockable panels for settings and history
• Persistent application settings
• Keyboard shortcuts for frequent actions
• Automatic model downloading from Hugging Face
• GPU (CUDA) support for fast translations

Backends

The application supports two backends:

1. GGUF (llama.cpp) - Recommended for most users

   • Uses quantized GGUF models
   • Lower memory usage
   • Good performance on CPU and GPU
   • Works on Windows

2. Transformers - For original models

   • Uses full precision models
   • Higher memory usage
   • Best quality
   • Works on Windows with CUDA

Requirements

• Python 3.10+ (due to transformers>=4.51.3 requirement)
• GPU with CUDA support (recommended) or CPU (slower)
• At least 8GB RAM (16GB recommended)
• About 5GB disk space for Q4_K_M model

Installation

Option 1: Automatic Installation (Recommended)

For Windows Command Prompt:

install.bat

For PowerShell:

.\install.ps1

Option 2: Manual Installation

1. Create virtual environment:

python -m venv venv

2. Activate virtual environment:

# Windows
venv\Scripts\activate

# Linux/Mac
source venv/bin/activate

3. Install required packages:

pip install -r requirements.txt

Usage

Quick Start

run.bat

Manual Start

# First activate virtual environment
venv\Scripts\activate

# Run application
python main.py

Model Download

The application will automatically detect and load available models. If you don't have any models:

1. Run the application
2. Click "Download Model"
3. Select a model (recommended: Q4_K_M - 4.6GB)
4. Wait for download to complete
5. Model will be automatically loaded

Available GGUF Models:

• Q4_K_M (4.6GB) - Recommended balance of quality/speed
• Q5_K_M (5.4GB) - Better quality, slower
• Q8_0 (8.0GB) - Best quality, requires more RAM

Original Model (for Transformers):

• Original Seed-X-PPO-7B (15GB) - Full precision, requires more VRAM

Supported Languages

• Arabic (ar), Czech (cs), Danish (da), German (de)
• English (en), Spanish (es), Finnish (fi), French (fr)
• Croatian (hr), Hungarian (hu), Indonesian (id), Italian (it)
• Japanese (ja), Korean (ko), Malay (ms), Norwegian Bokmål (nb)
• Dutch (nl), Norwegian (no), Polish (pl), Portuguese (pt)
• Romanian (ro), Russian (ru), Swedish (sv), Thai (th)
• Turkish (tr), Ukrainian (uk), Vietnamese (vi), Chinese (zh)

Project Structure

TranslatorSeedX/
├── main.py                     # Main entry point
├── run.bat                     # Windows batch file to run the application
├── install.bat                 # Installation script
├── install.ps1                 # PowerShell installation script
├── requirements.txt            # Python dependencies
├── README.md                   # Project documentation
├── .gitignore                  # Git ignore file
│
├── src/                        # Source code directory
│   ├── __init__.py
│   │
│   ├── gui/                    # GUI components
│   │   ├── __init__.py
│   │   ├── translator_app.py   # Main application window
│   │   └── filterable_combobox.py  # Custom combo box widget
│   │
│   ├── backend/                # Backend logic
│   │   ├── __init__.py
│   │   ├── translation_backend.py      # Translation manager and threads
│   │   ├── model_handler.py            # GGUF model handler (llama.cpp)
│   │   └── model_handler_transformers.py  # Transformers model handler
│   │
│   └── utils/                  # Utility modules
│       ├── __init__.py
│       ├── config.py           # Configuration settings
│       └── download_missing_files.py  # Model download utilities
│
└── models/                     # Model storage directory
    └── README.md               # Model information

Troubleshooting

Model Won't Load

• Check if you have enough RAM/VRAM
• Try a smaller model (Q4_K_M instead of Q8_0)
• Check console logs for error details
• For Transformers models: ensure CUDA is installed

Slow Translation

• Use GPU instead of CPU (check if CUDA is available)
• Reduce "Max Tokens" parameter in settings
• Use a smaller model
• Switch to Transformers backend for better GPU performance

CUDA Errors

• Ensure you have PyTorch with CUDA support installed
• Check if GPU drivers are up to date
• Try reinstalling PyTorch: pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

Contributing

We welcome contributions! Please see our Contributing Guide for details.

License

This project is licensed under the MIT License - see the LICENSE file for details.

The Seed-X-PPO-7B model is licensed under OpenMDW.

Acknowledgments

• ByteDance Seed Team for the Seed-X-PPO-7B model
• Mungert for GGUF quantization on Hugging Face
• llama.cpp for GGUF support
• Hugging Face for model hosting

Support

• 🐛 Report a Bug
• 💡 Request a Feature
• 💬 Discussions