Certification Exam: Custom activation functions


I was searching for this topic in the forum and until now I didn’t find an answer. Sorry if this is somewhere here, but I couldn’t find.

Now, let’s go with the question. I’ve planned to try the certification exam. I was digging into tensorflow in the last year, and I think i would be able to pass it. But one question arise when trying “simple” regression models. Could I use custom activation functions as, for example,…

tf_custom = lambda x: tf.math.pow(x,2)

Are these activations allowed during the exam?



I think you can use the custom activation functions while building your model in tensorflow certification exam. Because they evaluate your exam based on the model accuracy and saving your model.

Thank You!