MoE training for Me and You and maybe other people

Latent Collaboration in Multi-Agent Systems

A calm, CLI-native way to semantically grep everything, like code, images, pdfs and more.

[NeurIPS 2025] Scaling Speculative Decoding with Lookahead Reasoning

Mirage Persistent Kernel: Compiling LLMs into a MegaKernel

Nano vLLM

kernels, of the mega variety

Deep learning at the speed of light.

Big & Small LLMs working together

MAGI-1: Autoregressive Video Generation at Scale

A minimal PyTorch implementation of probabilistic diffusion models for 2D datasets.

Student version of Assignment 1 for Stanford CS336 - Language Modeling From Scratch

Async RL Training at Scale

🫀 Lovely Grad - tiny tensors need some love

HunyuanVideo: A Systematic Framework For Large Video Generation Model

Jupyter notebooks and other materials developed for the Columbia course APMA 4300

A Python Library for Learning Non-Euclidean Representations

Parallel Computing starter project to build GPU & CPU kernels in CUDA & C++ and call them from Python without a single line of CMake using PyBind11

A minimal GPU design in Verilog to learn how GPUs work from the ground up

Train an LLM to generate cracked Manim animations for mathematical concepts.

🍦 Never use print() to debug again.

A repository to unravel the language of GPUs, making their kernel conversations easy to understand

Kernels Written for 100 days of CUDA Challenge

Implementation of the sparse attention pattern proposed by the Deepseek team in their "Native Sparse Attention" paper

Master programming by recreating your favorite technologies from scratch.

Fully open reproduction of DeepSeek-R1

Apple AMX Instruction Set

Playing around "Less Slow" coding practices in C++ 20, C, CUDA, PTX, & Assembly, from numerics & SIMD to coroutines, ranges, exception handling, networking and user-space IO