Method step

torch/optim/sgd.py:56–92

Performs a single optimization step. Args: closure (Callable, optional): A closure that reevaluates the model and returns the loss.

(self, closure=None)

    @_use_grad_for_differentiable
    def step(self, closure=None):
        """Performs a single optimization step.

        Args:
            closure (Callable, optional): A closure that reevaluates the model
                and returns the loss.
        """
        loss = None
        if closure is not None:
            with torch.enable_grad():
                loss = closure()

        for group in self.param_groups:
            params_with_grad = []
            d_p_list = []
            momentum_buffer_list = []

            has_sparse_grad = self._init_group(group, params_with_grad, d_p_list, momentum_buffer_list)

            sgd(params_with_grad,
                d_p_list,
                momentum_buffer_list,
                weight_decay=group['weight_decay'],
                momentum=group['momentum'],
                lr=group['lr'],
                dampening=group['dampening'],
                nesterov=group['nesterov'],
                maximize=group['maximize'],
                has_sparse_grad=has_sparse_grad,
                foreach=group['foreach'])

            # update momentum_buffers in state
            for p, momentum_buffer in zip(params_with_grad, momentum_buffer_list):
                state = self.state[p]
                state['momentum_buffer'] = momentum_buffer

        return loss


SGD.__doc__ = r"""Implements stochastic gradient descent (optionally with momentum).