3 ms·
I was just thinking we didn’t have enough CUDA-accelerated numpy libraries. Jax, pytorch, vanilla TF, triton. They just don’t cut it
by whimsicalism 2y ago
I was just thinking we didn’t have enough CUDA-accelerated numpy libraries.
Jax, pytorch, vanilla TF, triton. They just don’t cut it