Hi, I was able to get the full marks for the assignment, but I am still learning PyTorch: I had a question on the use of retain_graph for the discriminator backward call in the provided code. disc_loss.backward(retain_graph=True) I checked the PyTorch documentation and forums to understand the pa…

Your First GAN assignment: use of retain_graph=True

Course Q&A Generative Adversarial Networks (GANS) Build Basic Generative Adversarial Networks

paulinpaloalto April 14, 2022, 5:06pm 2

Here’s another thread that’s worth a look, although I don’t think it directly addresses the role of retain_graph. It explains a couple of key points about how the gradients are managed.

Topic		Replies	Views
C1W1 Training Loop Build Basic Generative Adversarial Networks week-1	2	508	May 22, 2023
retain_graph=True? Build Basic Generative Adversarial Networks week-1	3	574	June 1, 2022
retain_graph=True and disc_opt.zero_grad() Build Basic Generative Adversarial Networks week-2 , week-3	6	805	December 17, 2022
Why do we set the retain_graph = True in the discriminators loss? Apply Generative Adversarial Networks week-3	1	590	November 5, 2021
Course 1 Week 1 Programming Assignment Build Basic Generative Adversarial Networks week-1	4	464	September 25, 2023

Your First GAN assignment: use of retain_graph=True

Related topics