karpathy

LLM Wiki

A pattern for building personal knowledge bases using LLMs.

This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.

The core idea

Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.

denji

Moved to git repository: https://github.com/denji/nginx-tuning

NGINX Tuning For Best Performance

For this configuration you can use web server you like, i decided, because i work mostly with it to use nginx.

Generally, properly configured nginx can handle up to 400K to 500K requests per second (clustered), most what i saw is 50K to 80K (non-clustered) requests per second and 30% CPU load, course, this was 2 x Intel Xeon with HyperThreading enabled, but it can work without problem on slower machines.

You must understand that this config is used in testing environment and not in production so you will need to find a way to implement most of those features best possible for your servers.