torch.nan_to_num_

torch.nan_to_num_(input, nan=0.0, posinf=None, neginf=None) → Tensor

In-place version of torch.nan_to_num()