torch.optim.lr_scheduler
ChainedScheduler
- get_last_lr - Get current learning rates.
- step - Perform a scheduler step on all chained schedulers.
- state_dict - Returns the state of all schedulers.
- load_state_dict - Loads the scheduler states.
CosineAnnealingWarmRestarts
- step - Perform a scheduler step.
CyclicLR
- get_last_lr - Get current learning rates.
- get_lr - Compute learning rates for current step.
- step - Perform a scheduler step.
- state_dict - Returns the state of the scheduler as a dictionary.
- load_state_dict - Loads the scheduler state.
LRScheduler
- get_last_lr - Get current learning rates.
- get_lr - Return the learning rate for each param group at the current epoch.
- step - Perform a scheduler step.
- state_dict - Returns the state of the scheduler as a dictionary.
- load_state_dict - Loads the scheduler state.
- print_lr - Print the learning rates for each param group.
OneCycleLR
- get_last_lr - Get current learning rates.
- get_lr - Compute learning rates for current step.
- step - Perform a scheduler step.
- state_dict - Returns the state of the scheduler as a dictionary.
- load_state_dict - Loads the scheduler state.
ReduceLROnPlateau
- get_last_lr - Get current learning rates.
- step - Perform a scheduler step.
- state_dict - Returns the state of the scheduler as a dictionary.
- load_state_dict - Loads the scheduler state.
SequentialLR
- get_last_lr - Get current learning rates.
- step - Perform a scheduler step.
- state_dict - Returns the state of all schedulers.
- load_state_dict - Loads the scheduler states.
Types
- ScaleFn - Scale function type
- ScaleMode - Scale mode
- LRLambda - Lambda function type
- SchedulerStateDict - Serialized scheduler state dictionary.
- LRSchedulerOptions - Options for LRScheduler constructor.
- StepOptions - Options for step method.
- PrintLrOptions - Options for print_lr method.
- AnnealStrategy - Anneal strategy
- PlateauMode - Mode for plateau detection