Fix no grad usage

[vasqu]

As per title, you need to "()" for the decorator to work as intended. Not sure about the implications but it certainly does not hurt to enable it and avoid any sort of unexpected memory surge

[vasqu]

Unsure about this one but I don't see a reason why we would not close here as well cc @remi-or

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[ArthurZucker]

Both are equivalent no?

[vasqu]

Both are equivalent no?

Did a sanity check

import torch
x = torch.tensor([1.], requires_grad=True)

def identity(x):
    return x * 1
z = identity(x)
z.requires_grad  # True

@torch.no_grad()
def doubler(x):
    return x * 2
z = doubler(x)
z.requires_grad  # False

@torch.no_grad
def tripler(x):
    return x * 3
z = tripler(x)
z.requires_grad  # False

You are correct, they are the same. I was confused about the standards around () - guess this is really nitpicky then. Will close this.