def clip_gradients(self, grad_clip_val: Union[int, float], optimizer: Optimizer, norm_type: float):
max_norm = grad_clip_val
norm_type = float(2.0)
optimizer = cast(OSS, optimizer)
optimizer.clip_grad_norm(max_norm, norm_type=norm_type)
def clip_grad_by_norm(self,
optimizer: Optimizer,
clip_val: Union[int, float],
norm_type: float = 2.0) -> None:
optimizer = cast(OSS, optimizer)
optimizer.clip_grad_norm(clip_val, norm_type=norm_type)