in middle of the notebook we created generated image by adding some noise to content image but just before training we intialised generated image to be the content image why is it so
That is a good point. They show how to create the initial generated image as the content image plus noise, but then they never use that output for anything. You’re right that before the training, they simply set the generated image to the content image before they start.
I will ask the course staff about this. Thanks for pointing out the issue!