Hi,
You have some more informations 'model.eval()' vs 'with torch.no_grad()') . Let me know if you have more questions.