In the week 1 Cats vs Dogs challenge, I had tweaked the following values:
[snippet removed by mentor]
Everything else was same
This gave me the following results:
Why did this happen??
Learning was back to normal(99% train and 80% val) when I used the default given values
Does this help?
Model analysis is covered in good level of detail in deep learning specialization. Please complete it before moving forward.
Not really. My code is running without errors. Based on the thread you pointed to, my filters were in the power of 2 and the learning rate wasn’t too big or small.
What’s bothering me is that the accuracy remained the same for all 15 epochs. Did I hit a plateau?
Im just looking for an explanation for this. I enroll in the Deep learning course after I finish this one.
If you are using the correct activation function for the final layer, please click my name and message your notebook as an attachment.
I had the same thing happen the first time I ran it, its a problem with the meta data, I had to shut it down and restart from the beginning
As far as assignments in this specialization are concerned, notebook metadata has to do with editing / removing cells and helping the grader evaluate your assignment properly. Restarting the kernal won’t change notebook metadata.
Restarting and running the notebook runs all cells in the right order exactly once and this got the job done for you.
