HybrIE CLI

HybrIE CLI - Command-line interface for HybrIE (Hybrid Inference Engine for Heterogeneous Hardware to Power Generative AI Models)

🚀 Quick Start

Installation

Install globally via npm:

npm install -g @tosi-n/hybrie-cli

Usage

# Show help
hybrie --help

# Generate an image
hybrie img-gen "a beautiful sunset over mountains"

# Start the server
hybrie-server --help

📋 Features

• Cross-platform: Works on macOS and Linux
• Multiple architectures: Supports x64 and ARM64
• FLUX model support: Generate high-quality images with FLUX models
• Local inference: Run models locally on your hardware
• Server mode: Run as a server for remote access
• Model management: Download and manage models automatically
• Beautiful UI: Rich terminal interface with progress bars and colors

🖥️ Supported Platforms

| Platform | Architecture | Status | |----------|-------------|--------| | macOS | ARM64 | ✅ | | Linux | x64 | ✅ |

📖 Commands

Image Generation

# Basic generation
hybrie img-gen "a cat sitting on a rainbow"

# With specific model
hybrie img-gen "a futuristic city" --model flux.1-dev

# With custom settings
hybrie img-gen "a dragon" --steps 50 --size 1024x1024

# Save to specific directory
hybrie img-gen "a sunset" --output ./my-images

Model Management

# List available models
hybrie models list

# Download a model
hybrie models download flux.1-dev

# Show model info
hybrie models info flux.1-schnell

Server Mode

# Start server
hybrie-server start

# Start with custom config
hybrie-server start --config ./my-config.toml

# Check server status
hybrie-server status

⚙️ Configuration

HybrIE CLI automatically creates configuration files in:

• macOS/Linux: ~/.config/hybrie/

Configuration File (hybrie-config.toml)

[server]
host = "127.0.0.1"
port = 8080
max_concurrent_requests = 4
request_timeout_seconds = 300
enable_reflection = true

[models]
cache_dir = "~/.cache/hybrie/models"
auto_download = ["flux.1-schnell"]
max_cache_size_gb = 50.0
preload_models = []

[inference]
device_preference = "auto"  # auto, cpu, metal, cuda
batch_size = 1
max_steps = 50

[api]
enable_cors = true
api_key_required = false
rate_limit_per_minute = 60

🔧 Advanced Usage

Environment Variables

# Set log level
export HYBRIE_LOG_LEVEL=debug

# Set custom cache directory
export HYBRIE_CACHE_DIR=/path/to/cache

# Set device preference
export HYBRIE_DEVICE=metal  # or cuda, cpu

Custom Models

# Use local model directory
hybrie img-gen "a test image" --model-path /path/to/model

# Use HuggingFace model
hybrie img-gen "a test image" --model black-forest-labs/FLUX.1-dev

🐛 Troubleshooting

Common Issues

Binary not found after installation:

# Re-run the installer
npm install -g @tosi-n/hybrie-cli --force

Permission denied on macOS/Linux:

# Make sure the binary is executable
chmod +x $(which hybrie)

Model download fails:

# Check internet connection and try again
hybrie models download flux.1-schnell --force

Getting Help

• Documentation: https://github.com/hybrie/hybrie-cli
• Issues: https://github.com/hybrie/hybrie-cli/issues
• Discussions: https://github.com/hybrie/hybrie-cli/discussions

🔄 Updates

Update to the latest version:

npm update -g @hybrie/cli

📄 License

This project is licensed under the MIT OR Apache-2.0 License - see the LICENSE file for details.

🤝 Contributing

Contributions are welcome! Please see our Contributing Guide for details.

HybrIE CLI - Making AI inference accessible across all platforms 🚀