Adds the `no_grad` context manager to `Tensor` and `Module`.
The `no_grad` context manager can be used to temporarily disable gradient computations for a `Tensor` or `Module`. It can be used to speed up computations in situations where `backward` will not be called, such as when evaluating on a test or validation set.
python
with tensor.no_grad():
...
python
with module.no_grad():
...