Inference Training with Retrieval Augmented Generation - EXAKI training

EXAKI training number 2.

This webinar, led by METU, introduces the EXA4MIND AI Inference Service, which supports advanced querying capabilities in high-performance computing (HPC) environments.

During the training session, a local implementation of Retrieval-Augmented Generation (RAG) is showcased, demonstrating how it can enhance the precision of large language model (LLM) responses.

The session walks through a complete RAG pipeline—from preprocessing a university knowledge base into manageable chunks, generating vector embeddings using an embedding model, to storing these embeddings in a vector database for semantic similarity search.

The training concludes by demonstrating how to send an API request to the inference service using the enriched prompt.