lr_scheduling_policy
LRSchedulingPolicy
Defines a learning rate scheduling policy.
Source code in lightning_boost/modules/optim/lr_scheduling_policy.py
__init__(interval='epoch', frequency=1, monitor='val_total-loss')
Initializes LR scheduling policy.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
interval |
str
|
Interval in which LR scheduling is performed ('step' or 'epoch'). Defaults to 'epoch'. |
'epoch'
|
frequency |
int
|
Frequency in which LR scheduling is performed. Defaults to 1. |
1
|
monitor |
str
|
Monitor metric. Defaults to 'val_total-loss'. |
'val_total-loss'
|
Source code in lightning_boost/modules/optim/lr_scheduling_policy.py
bind_lr_scheduler(lr_scheduler)
Binds learning rate scheduler and returns scheduling policy.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
lr_scheduler |
LRScheduler
|
Learning rate scheduler. |
required |
Returns:
Type | Description |
---|---|
Dict[str, Any]
|
Dict[str, Any]: LR scheduling policy. |