Project Title: AD 331 Implementing a Simple Retrieval-Augmented Generation (RAG) System

By: Amde Wubshet

LLM Used:

GPT-2

Embedding Model Used

all-MiniLM-L6-v2

Test Cases

Test Case	Chunks Recieved	Query	Analysis
1	Chunk 2	"What is the primary function of the Baku?"	The knowledge chunk returned is what was expected, although the LLM still ended up hallucinated
2	Chunk 1	"How are you supposed to cook sausages?"	What was returned from the LLM and the RAG system was not as ideal since the question isn't in the KB.
3	Chunk 2, Chunk 3	"Which creature has a powerful sense of smell, and which creature is known for consuming dreams?"	Top 2 chunks ended up as expected, although the LLM is still hallucinating.