simonguo-cohere's Stars
HazyResearch/ThunderKittens
Tile primitives for speedy kernels
stanford-crfm/levanter
Legible, Scalable, Reproducible Foundation Models with Named Tensors and Jax
abhinavgoel95/collective-matmul-unit-tests
google/paxml
Pax is a Jax-based machine learning framework for training large scale models. Pax allows for advanced and fully configurable experimentation and parallelization, and has demonstrated industry leading model flop utilization rates.
NVIDIA/JAX-Toolbox
JAX-Toolbox
karpathy/llm.c
LLM training in simple, raw C/CUDA
rwitten/HighPerfLLMs2024