Chat with your data. GINA turns plain-English questions into grounded answers by running real SQL on your dataset — not guessing from model weights. Built with Next.js, Fastify, Groq, and pgvector.
Table-Augmented Generation (TAG) in LangChain: guarded read-only SQL + row-wise LM semantic operators + answer generation, as a pipeline and a create_agent ReAct agent. Runs 100% locally on ollama.
To associate your repository with the table-augmented-generation topic, visit your repo's landing page and select "manage topics."