Opiniated RAG for integrating GenAI in your apps 🧠Focus on your product rather than the RAG. Easy integration in existing products with customisation! Any LLM: GPT4, Groq, Llama. Any Vectorstore: PGVector, Faiss. Any Files. Anyway you want.
Today, if a user input contains different questions, all questions are treated as a single one, hence the retrieved chunks might or might not contain the information necessary to answer the different questions.
With this new feature, the system will automatically decompose the user input into multiple (self-contained) questions, which are treated independently in the retrieval step. This enables the system to retrieve chunks relevent to each question. The answer generation phase then considers all retrieved chunks, allowing the system to generate relevant answers even for user inputs containing different questions.
linear[bot]
commented
3 weeks ago
Today, if a user input contains different questions, all questions are treated as a single one, hence the retrieved chunks might or might not contain the information necessary to answer the different questions.
With this new feature, the system will automatically decompose the user input into multiple (self-contained) questions, which are treated independently in the retrieval step. This enables the system to retrieve chunks relevent to each question. The answer generation phase then considers all retrieved chunks, allowing the system to generate relevant answers even for user inputs containing different questions.