mindspore.experimental.optim.lr_scheduler.LambdaLR

class mindspore.experimental.optim.lr_scheduler.LambdaLR(optimizer, lr_lambda, last_epoch=- 1)[source]

Sets the learning rate of each parameter group to the initial lr times a given function. When last_epoch=-1, sets initial lr as lr.

Warning

This is an experimental lr scheduler module that is subject to change. This module must be used with optimizers in Experimental Optimizer .

Parameters

optimizer (mindspore.experimental.optim.Optimizer) – Wrapped optimizer.
lr_lambda (Union(function, list)) – A function which computes a multiplicative factor given a parameter last_epoch, or a list of such functions, one for each group in optimizer.param_groups.
last_epoch (int, optional) – The index of the last epoch. Default: -1.

Raises

ValueError – If the length of lr_lambda is not equal to the number of param groups.

Supported Platforms:: Ascend GPU CPU

Examples

>>> from mindspore import nn
>>> from mindspore.experimental import optim
>>> net = nn.Dense(2, 3)
>>> optimizer = optim.Adam(net.trainable_params(), 0.01)
>>> lmbda = lambda epoch: 0.9 ** epoch
>>> scheduler = optim.lr_scheduler.LambdaLR(optimizer, lr_lambda=[lmbda])
>>> for i in range(3):
...     scheduler.step()
...     current_lr = scheduler.get_last_lr()
...     print(current_lr)
[Tensor(shape=[], dtype=Float32, value= 0.009)]
[Tensor(shape=[], dtype=Float32, value= 0.0081)]
[Tensor(shape=[], dtype=Float32, value= 0.00729)]