Large Language Models Integration

The SBA Framework LLM Integration is available in a separate branch chatgpt-integration of the SBA respository.

Please read the documentation of the Sinequa Retrieval Augmented Generation repository for guidance on how to set up the necessary plugins and configuration on your Sinequa server.

This application is a customization of Vanilla Search in which a chat component (sq-chat from @sinequa/components/machine-learning) is used to interact with a Large-Language Model (LLM), such as GPT-4, the model powering the popular ChatGPT.

This application is not meant to be used as-is, but rather as a set of examples to understand how to integrate a LLM in a Sinequa application for various use cases:

• Retrieval Augmented Generation (RAG): the LLM is fed with search results and prompted to generate an answer to the user query, along with a summary of the documents.
• Query intent detection: the LLM takes a search query from the user, rephrases it and automatically applies filters.
• Document summarization: the LLM summarizes a set of passages extracted from a document.
• Entity extraction: the LLM generates a graph of entities from a set of passages extracted from a document.
• Translation: the LLM translates a search query from one language to another.
• Personnalized greeting: the LLM generates a greeting message based on the user's information.