Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
Making a local, persistent and self-curated wiki with the Andrej Karpathy knowledge base
Learn to build a self-curated wiki using Markdown and LLMs, a RAG approach without vector databases, for persistent and clean knowledge management.
The dream for every developer is to have a custom knowledge base (from a technology, internal procedures or just personal knowledge mapping), inputting the information is easy, integrating it to the knowledge base is hard but keeping the references without inconsistencies is where most of the people drops the idea. Andrej Karpathy proposed the LLM Wiki pattern, a way to create a knowledge base with the detail of using an LLM as the knowledge base curator, this way, the LLMs can clean the data and solve inconsistencies as the references grow
This approach doesn’t use vectorial DB to ingest data, instead, uses Markdown text and an index for pointing where the information is
Here I will show a custom implementation of that pattern only using skills for leveraging knowledge understand on multiple levels