nsl.configs.DecayConfig

View source on GitHub

Contains configuration for decaying a value during training.

decay_steps A scalar int32 or int64 Tensor or a Python number that specifies the decay frequency, specied in units of training steps. Must be positive.
decay_rate A scalar float32 or float64 Tensor or a Python number. Defaults to 0.96.
min_value minimal acceptable value after applying decay. Defaults to 0.0.
decay_type Type of decay function to apply. Defaults to nsl.configs.DecayType.EXPONENTIAL_DECAY.

Methods

__eq__

Return self==value.

__ge__

Automatically created by attrs.

__gt__

Automatically created by attrs.

__le__

Automatically created by attrs.

__lt__

Automatically created by attrs.

__ne__

Check equality and either forward a NotImplemented or return the result negated.