Linear layer input neurons number calculation after conv2d

vision
10

You could also use torch.nn.LazyLinear, a module where in_features is inferred.
Please refer to Inferring shape via flatten operator - #20 by iacob

2 Likes