Project
RAG + LLM Answering v6
A local retrieval-augmented generation experiment exploring custom retrieval, context prompting, and LLM-based question answering.
Explore project01
Overview
The project explores how retrieved context can be combined with a local language model to build a more focused question-answering system.
02
Problem
Language models can produce unreliable answers when they lack relevant context or access to the right information.
03
Approach
The system experiments with custom retrieval, context prompting, and MLX-LM to provide relevant context to a local language model before generating an answer.
04
Results
Answerable Questions
Passed
Unanswerable Questions
Passed
Retrieval Score
0.4842
Stack
Technologies
PythonRAGLLMsMLX-LM
Explore