I am just curious to know, how does PyTorch track operations on tensors (after the .requires_grad is set as True and how does it later calculate the gradients automatically. Please help me understand the idea behind autograd. Thanks. Continue reading...
