Stars
AI agents running research on single-GPU nanochat training automatically
🚀 A very efficient Texas Holdem GTO solver
Use your Neovim like using Cursor AI IDE!
Lightweight coding agent that runs in your terminal
Neural Networks: Zero to Hero
tiktoken is a fast BPE tokeniser for use with OpenAI's models.
Entropy Based Sampling and Parallel CoT Decoding
g1: Using Llama-3.1 70b on Groq to create o1-like reasoning chains
LlamaIndex is the document processing platform for AI