The flatten () function takes in a tensor t as an argument. This is how Torchvision's CNN implementations deal with variable size image inputs.
確定! 回上一頁