fastRAG: Efficient Retrieval Augmentation and Generation Framework
fastRAG is a research framework designed for building and exploring efficient retrieval-augmented generative models and applications. It incorporates state-of-the-art large language models (LLMs) and information retrieval techniques, empowering researchers and developers with a comprehensive toolset for advancing retrieval-augmented generation.
Key Features:
- Compatibility: Now compatible with Haystack v2+.
- Optimized for Intel Hardware: Leverage Intel extensions for PyTorch (IPEX) and other optimizations for running efficiently on IntelĀ® hardware.
- Customizable: Built using Haystack and HuggingFace, ensuring all components are 100% Haystack compatible.
- Installation: Easy setup via pip with additional packages for specific use cases.
Benefits:
- Research Empowerment: Provides tools for researchers to build and test generative models effectively.
- Performance: Optimized for compute efficiency, making it suitable for large-scale applications.
- Community Engagement: Open to feedback, suggestions, and contributions from the community.