Views
No views yet
1import torch
2from kernels import get_kernel
3# Download optimized kernels from the Hugging Face hub
4ops = get_kernel("mohitsha/aiter_aiter_operator")
5# Random tensor
6x = torch.randn((10, 10), dtype=torch.bfloat16, device="cuda")
7y = torch.randn((10, 10), dtype=torch.bfloat16, device="cuda")
8out = ops.div(y, x)
9print(out)
10