model-loading
Drop-in Rust replacement for safetensors that loads model weights faster
Hardware Control GateKeeper Kernels for AI inference within frameworks.
CUDA 12 accelerated backend for safetensors-streaming