Implementation about a recommender System using RQ-VAE Semantic IDs

Replication code for semantic ID generation in “Transformer Memory as a Differentiable Search Index”.

Semantic ID Generation based on RQ-VAE Architecture for Recommendation purposes

A production-ready Retrieval-Augmented Generation (RAG) API featuring multi-format document support (PDF/DOCX/TXT), streaming LLM responses, bilingual output (ID/EN), and semantic vector search. Built with FastAPI, LangChain, PostgreSQL + pgvector, and Groq.