Explore live open-source projects and AI models.

No description available from source.

A notebook based tutorial series on buildling a LLM from scratch

ToyLLM: Learning LLM from Scratch

Building an LLM inference engine from scratch

The GPU-free LLM inference engine. Combines lazy expert loading + TurboQuant KV compression to run models that shouldn't fit on your hardware. Built from scratch, fully local, zero cloud.

RepoZero: Can LLMs Generate a Code Repository from Scratch? (https://arxiv.org/abs/2605.07122)

🔬 The most atomic GPT-2 implementation in 265 lines of pure Python & CUDA. A bilingual "Rosetta Stone" for understanding LLM internals from scratch. No dependencies, just math and kernels.

Implement and train a Tiny LLM from scratch!

Trained a 114 million Parameter LLM from Scratch.

Chat with Documents from scratch using LLMs and a vector databse