Can we use the Adam optimizer?

Course Q&A TensorFlow Developer Professional Convolutional Neural Networks in TensorFlow
1

On the C2W2 lab, I figured I’d try to use the Adam optimizer to speed things up since the training is super slow and requires babysitting or it self-cancels. However, when I tried to do so using similar syntax as RMSprop, it threw an error. In particular, this works:

  from tensorflow.keras.optimizers import RMSprop
  model.compile(optimizer=RMSprop(learning_rate=0.001),
                loss='binary_crossentropy',
                metrics=['accuracy'])

But this throws an error. What am I doing wrong?

  from tensorflow.keras.optimizers import Adam
  model.compile(optimizer=Adam(learning_rate=0.001),
                loss='binary_crossentropy',
                metrics=['accuracy'])
2

What you need to do is go to tensroflow website and check how to use the adam optimizer and this is the guide of how you do it in general.

3

Thanks for the reply. Checking the documentation was the first thing I tried before posting, and I observed that at a high level, the two optimizers’ documentation suggested identical syntax. I also checked my DLS C2 notes which covered Adam, but they did not use it in TensorFlow. If you or others have a more specific hint or can point to an example similar to those used in the course, that would be welcome. Thanks again.

4

Changing from RMSProp to Adam works as expected… for me. Is it any “formal” error (syntax or whatsoever) you encounter?