RuntimeError: memory format option is only supported by strided tensors

[eran88]

Im only using the activation in a simple model:
class KAT(nn.Module):
def init(
self,

) -> None:
    super().__init__()
    self.act1 = KAT_Group(mode="identity")

def forward(self, x: Tensor) -> Tensor:
    x = self.act1(x)

and getting the following error:

x = kat_rational_cu.rational_fwd_1dgroup(input, weight_numerator, weight_denominator, group)
RuntimeError: memory format option is only supported by strided tensors

[Adamdad]

Could you please provide more context about the problem? For example, the environment version, system details, and any other relevant information. Also, are you using ROCm devices?

[eran88]

PyTorch Version: 2.2.0+cu118
CUDA Available: True
CUDA Version: 11.8
Number of CUDA devices: 8
CUDA Device Name: NVIDIA A100-SXM4-80GB
ROCm Available: False
Python Version: 3.10.9