Course Introduction
This episode compares two newer RAG approaches. HiRAG reduces noise and hallucination through hierarchical retrieval, selection, composition, and reasoning, while HeteRAG aligns retrieval and generation through heterogeneous embedding spaces. The workflow and code clarify when each improves on conventional vector search.
Total video duration: 23 min 24 sec.
Companion Git repository: https://qytgit.qytang.com/qytadmin/2025-26-HiRAG-HeteRAG.
COURSE RECORDINGS
Course Video
Watch the full recording here or open it on the original video platform.
BILIBILI01