Self-Retrieval proposes an end-to-end, LLM-driven information retrieval architecture that fully internalizes the required abilities of IR systems into a single LLM and leverages the capabilities of LLMs during the IR process.
Self-Retrieval proposes an end-to-end, LLM-driven information retrieval architecture that fully internalizes the required abilities of IR systems into a single LLM, significantly enhancing retrieval performance and enabling retrieval augmented generation.