TurboPilot

Runs entirely locally on your machine, eliminating cloud dependencies and ensuring code privacy, as emphasized in the project's philosophy and local execution feature.

Supports various state-of-the-art models like StableCode, WizardCoder, and StarCoder, allowing customization based on hardware and language needs, detailed in MODELS.md.

Offers both CPU and GPU inference via CUDA, enabling performance tuning for different setups, as shown in the CUDA Docker and executable instructions.

Works with the FauxPilot plugin for inline completions, providing a Copilot-like experience directly in the editor, as demonstrated in the setup guide.