For all of the 28x28 pixel greyscale images in the datasets the image size is 784 and so the output tensor for the training set images is of shape [55000, 784] ...
確定! 回上一頁