Hello everyone. I’ve been attempting to adapt * lesson7-superres-gan.ipynb to work on a different dataset. I’m very new and still trying to grasp the basics. Any idea why this might be happening?:
The main thing I’m concerned about is the labels having different columns than the data. Ideally I’d have : epoch,train_loss, gen_loss, disc_loss, and time. I had to tweak several parts of the code to adapt to the folder structure I was using, but I’m not sure where to look.
If you have any insights into the numbers I’m getting, I’m interested in that as well. They’re nowhere in the range of the ones from the lesson, but it was getting interesting results in the pretraining steps. It hasn’t completed the first gan fit yet.
One last question - If I stop in the middle of the fit, would it be partially fitted, corrupted, or unchanged?
Thanks for any help. I only discovered this whole thing a few days ago. It’s a life-changing resource.