Welcome to LocalRQA’s documentation!#
LocalRQA is an open-source toolkit that enables researchers and developers to easily train, test, and deploy retrival-augmented QA (RQA) systems using techniques from recent research. Given a collection of documents, you can use pre-built pipelines in our framework to quickly assemble an RQA system using the best off-the-shelf models. Alternatively, you can create your own training data, train open-source models using algorithms from the latest research, and deploy your very own local RQA system!
Components in LocalRQA#
🌟 Why LocalRQA?#
LocalRQA is a toolkit designed to make researching and developing retrieval-augmented QA systems more efficient and effective. It offers a variety of training methods curated from latest research (see Training) and automatic evaluation metrics (see Evaluation) to help users develop new RQA approaches and compare with prior work.
Moreover, LocalRQA doesn’t just stop at creating these systems; it also provides tools for deploying these systems and improving them through real-world feedback (see Serving). We offer support with popular inference acceleration frameworks such as vLLM and SGLang, as well as methods to directly launch your RQA system with an interactive UI to allow users to chat with your system and provide feedback!
With a comprehensive suite of tools, LocalRQA aims to make it easier for researchers and developers to train, test, and deploy novel RQA approaches!
🚀 Getting Started!#
To get started with using LocalRQA, please refer to our Installation and Quickstart guide!