Hi, I am getting an error when trying to use max.ops.conv2d with non-unit dilation:
ValueError: Non-unit dilation is not supported yet.
This happens during compilation when calling max.ops.conv2d() with dilation=(2, 2) or any dilation beyond the default.
Interestingly, max.ops.conv2d_transpose() works fine with non-unit dilation.
I could not find any related warnings/error messages in the open-sourced max engine api in Python or the docs, nor in the Mojo kernel impls.
Is this a known issue, or am I doing sth wrong?
Looks like a bug. Could you please create an issue?
1 Like