What is the difference between model.eval() and torch.no_grad()?

eval() changes layer behavior, while no_grad() disables gradient computation.

model.eval() affects layers like dropout and batchnorm. torch.no_grad() disables autograd tracking. They are often used together but serve different purposes.