C3W2 assignment model accuracy way too low with multiple architectures

In the end when I’m training the model, I keep getting an accuracy of 0.227 no matter what architecture I use. Is there something I’m missing or a workaround?

1 Like


  1. Loss function is correct
  2. Number of neurons in output layer matches the number of classes in the dataset.

Please click my name and message your notebook as an attachment.