Trefolio Learning

Hands-on, runnable demos of AI/ML techniques

Every demo here is real, runnable code, not slides — the point is to see a technique work (and sometimes fail) step by step, with the actual source browsable inline, not just read a description of it.

RAG Live

A real, runnable Retrieval-Augmented Generation pipeline: chunking, embeddings, vector search, prompt assembly, and streamed generation — watched live, step by step. Includes a Classic vs. Agentic RAG comparison, a measured Embeddings vs. keyword-search comparison, and real Recall@K / Faithfulness evaluation metrics.

Embeddings Vector search Agentic RAG Evaluation
Open the demo →

More techniques will be added here over time.