#assigment transfer learning with mobilenet
#specialization
#week2
Convolutional Neural Networks
Hi
Why in the second assignment in weak 2 (convolutionneural network) we use linear activation?
We have only 2 class Alpaca/Not Alpaca
So why does it work correct?
1 Like
Right, the whole key point here is “Use from_logits=True” which is basically applying a sigmoid to the output!
Check this page for more info:
https://www.tensorflow.org/api_docs/python/tf/keras/losses/BinaryCrossentropy
1 Like
Ty so much 
1 Like