It took me a long time to understand that the input to happyModel should just be a list of layers which are instances of tf.keras.layers. I think this should be explained explicitly since there are many different ways to create models in TF. I only guessed this because of the [ at the beginning, but that was more luck than any logic.
Another difficulty is understanding the tensorflow documentation, which is not completely clear either. For example, it doesn’t include input_shape as an input parameter in the API documentation and I had to figure this out from reading other comments here in the discussion, so thought I write a note here for anyone else who got stuck.
1 Like
PS. Looking back, it seems a bit clearer. The documentation for tf.keras.Sequential() has this info in the definition. However, it could be made easier to find the correct formatting.
1 Like
@StuHaze the comments embedded in the notebook really disguise what is expected in that code block. Glad you got past it.
For others who are still dealing with this issue, this related thread might help…
[Tips for troubles with Sequential and Functional API syntax]
1 Like