I am working on federated learning, the dataset we have comprises 5 classes (0, 1, 2, 3, 4), distributed among 10 clients. To simulate a realistic scenario, not all clients have all the classes. For example, in client 1, only classes (0, 1, 2, 3) are present, and so forth.
The encountered errors:
target.shape.assert_is_compatible_with(output.shape)
ValueError: Shapes (None, 4) and (None, 5) are incompatible.
Could you please advise me to solve this error?