Hi,
I’ve created a custom linear layer and am now testing it by running the forward() function on tensors of different sizes. I’m finding that for certain dimensional sizes, the forward pass throws a size mismatch error. For example, running forward on a (2, 10) tensor (and using 2 input features and 10 output features) works properly, while running forward on a (3, 10) tensor with 3 input features and 10 output features does not work.
Any insight on what to check for would be greatly appreciated.
Thanks!
Edit: Issue resolved, made a typo initializing the tensor sizes I was testing with.