← back
complete
GBTE RAG Search
Ask a tennis question and receive an answer based in GreatBase content. Built on 300+ hours of GBTE's own podcasts and courses.
About
link to repoRetrieval augmented generation (RAG) pairs a search step with an LLM's generation step. The model answers based mainly on context retrieved from a specific, private data source, not what it memorized in training.
GBTE's content lives inside video courses and podcasts general models were never trained on. RAG grounds every answer in the unseen data, with a citation to the exact moment it was said.
24,486
embedded chunks
300+
podcast episodes
31+ hrs
video courses
Pipelines
IngestionIngestion | data to embeddings
QueryQuery | question to answer
Ingestion | data to embeddings
Query | question to answer
Live demo
Notes
▾