ValueError: Shapes (None, 1) and (None, 3) are incompatible

This is from the TF1 Course 2 (Convolutional Neural Networks in TensorFlow). Please move this to the correct category.

As Paul said, ‘use the “sparse” version of categorical_crossentropy loss’. Also, you are passing too many Arguments to ImageDataGenerator. I think that is not necessary. Moreover, your parse_data_from_input code seems incorrect to me.

Please move this to the correct category and wait for the official mentors of this course to respond to you.