Gen-AI Developer Classroom notes 12/Apr/2025

RAG

• Overview
  

• If we have to build a RAG, We need to deal with

  • different datasources
  • different chunking techniques
  • different embeddings (Transformers/models)
  • different vector databases
  • different generators
• Refer Here for the sequence of steps

• NOte: This RAG is ineffective as it uses models which are small and can be run on laptops

• To simplify the above interactions, We have a framework called as langchain which includes all the above mentioned stuff

  • Refer Here for langchain components