Python TorchPolicy.extra_grad_process Examples

Example #1

File: torch_policy_template.py Project: priyadwivedi20/ray

        def extra_grad_process(self, optimizer, loss):
            """Called after optimizer.zero_grad() and loss.backward() calls.

            Allows for gradient processing before optimizer.step() is called.
            E.g. for gradient clipping.
            """
            if extra_grad_process_fn:
                return extra_grad_process_fn(self, optimizer, loss)
            else:
                return TorchPolicy.extra_grad_process(self, optimizer, loss)

Example #2

 def extra_grad_process(self):
     if extra_grad_process_fn:
         return extra_grad_process_fn(self)
     else:
         return TorchPolicy.extra_grad_process(self)