pyiqa.models.lr_scheduler
=========================

.. py:module:: pyiqa.models.lr_scheduler






Module Contents
---------------

.. py:class:: MultiStepRestartLR(optimizer, milestones, gamma=0.1, restarts=(0, ), restart_weights=(1, ), last_epoch=-1)

   Bases: :py:obj:`torch.optim.lr_scheduler._LRScheduler`


   MultiStep with restarts learning rate scheme.

   :param optimizer: Torch optimizer.
   :type optimizer: torch.nn.optimizer
   :param milestones: Iterations that will decrease learning rate.
   :type milestones: list
   :param gamma: Decrease ratio. Default: 0.1.
   :type gamma: float
   :param restarts: Restart iterations. Default: [0].
   :type restarts: list
   :param restart_weights: Restart weights at each restart iteration.
                           Default: [1].
   :type restart_weights: list
   :param last_epoch: Used in _LRScheduler. Default: -1.
   :type last_epoch: int


   .. py:method:: get_lr()


.. py:function:: get_position_from_periods(iteration, cumulative_period)

   Get the position from a period list.

   It will return the index of the right-closest number in the period list.
   For example, the cumulative_period = [100, 200, 300, 400],
   if iteration == 50, return 0;
   if iteration == 210, return 2;
   if iteration == 300, return 2.

   :param iteration: Current iteration.
   :type iteration: int
   :param cumulative_period: Cumulative period list.
   :type cumulative_period: list[int]

   :returns: The position of the right-closest number in the period list.
   :rtype: int


.. py:class:: CosineAnnealingRestartLR(optimizer, periods, restart_weights=(1, ), eta_min=0, last_epoch=-1)

   Bases: :py:obj:`torch.optim.lr_scheduler._LRScheduler`


   Cosine annealing with restarts learning rate scheme.

   An example of config:
   periods = [10, 10, 10, 10]
   restart_weights = [1, 0.5, 0.5, 0.5]
   eta_min=1e-7

   It has four cycles, each has 10 iterations. At 10th, 20th, 30th, the
   scheduler will restart with the weights in restart_weights.

   :param optimizer: Torch optimizer.
   :type optimizer: torch.nn.optimizer
   :param periods: Period for each cosine anneling cycle.
   :type periods: list
   :param restart_weights: Restart weights at each restart iteration.
                           Default: [1].
   :type restart_weights: list
   :param eta_min: The minimum lr. Default: 0.
   :type eta_min: float
   :param last_epoch: Used in _LRScheduler. Default: -1.
   :type last_epoch: int


   .. py:method:: get_lr()