ChainedScheduler — PyTorch 2.7 documentation (original) (raw)

class torch.optim.lr_scheduler.ChainedScheduler(schedulers, optimizer=None)[source][source]

Chains a list of learning rate schedulers.

Takes in a sequence of chainable learning rate schedulers and calls their step() functions consecutively in just one call to step().

Parameters

Example

Assuming optimizer uses lr = 1. for all groups

lr = 0.09 if epoch == 0

lr = 0.081 if epoch == 1

lr = 0.729 if epoch == 2

lr = 0.6561 if epoch == 3

lr = 0.59049 if epoch >= 4

scheduler1 = ConstantLR(optimizer, factor=0.1, total_iters=2) scheduler2 = ExponentialLR(optimizer, gamma=0.9) scheduler = ChainedScheduler([scheduler1, scheduler2], optimizer=optimizer) for epoch in range(100): train(...) validate(...) scheduler.step()

get_last_lr()[source]

Return last computed learning rate by current scheduler.

Return type

list[float]

get_lr()[source]

Compute learning rate using chainable form of the scheduler.

Return type

list[float]

load_state_dict(state_dict)[source][source]

Load the scheduler’s state.

Parameters

state_dict (dict) – scheduler state. Should be an object returned from a call to state_dict().

state_dict()[source][source]

Return the state of the scheduler as a dict.

It contains an entry for every variable in self.__dict__ which is not the optimizer. The wrapped scheduler states will also be saved.

step()[source][source]

Perform a step.