Closed ltindall closed 6 years ago
When changing the input to the first convolutional layer the output of the layer became too small for the second convolutional layer.
output width = ((input width - filter size + 2*padding)/stride )+ 1
When changing z input to size 3 the output width becomes 2. ((3-3+2*1)/2+1) = 2
So the second convolutional layer filter must have z size <= 2.
When changing the input to the first convolutional layer the output of the layer became too small for the second convolutional layer.
output width = ((input width - filter size + 2*padding)/stride )+ 1
When changing z input to size 3 the output width becomes 2. ((3-3+2*1)/2+1) = 2
So the second convolutional layer filter must have z size <= 2.