TensorFlow 2.0 Beta is available Learn more

tf.keras.experimental.CosineDecayRestarts

TensorFlow 2.0 version View source on GitHub

Class CosineDecayRestarts

A LearningRateSchedule that uses a cosine decay schedule with restarts.

Inherits From: LearningRateSchedule

Aliases:

  • Class tf.compat.v1.keras.experimental.CosineDecayRestarts
  • Class tf.compat.v2.keras.experimental.CosineDecayRestarts
  • Class tf.keras.experimental.CosineDecayRestarts

__init__

View source

__init__(
    initial_learning_rate,
    first_decay_steps,
    t_mul=2.0,
    m_mul=1.0,
    alpha=0.0,
    name=None
)

Applies cosine decay with restarts to the learning rate.

See [Loshchilov & Hutter, ICLR2016], SGDR: Stochastic Gradient Descent with Warm Restarts. https://arxiv.org/abs/1608.03983

When training a model, it is often recommended to lower the learning rate as the training progresses. This schedule applies a cosine decay function with restarts to an optimizer step, given a provided initial learning rate. It requires a step value to compute the decayed learning rate. You can just pass a TensorFlow variable that you increment at each training step.

The schedule a 1-arg callable that produces a decayed learning rate when passed the current optimizer step. This can be useful for changing the learning rate value across different invocations of optimizer functions.

The learning rate multiplier first decays from 1 to alpha for first_decay_steps steps. Then, a warm restart is performed. Each new warm restart runs for t_mul times more steps and with m_mul times smaller initial learning rate.

Example usage:

first_decay_steps = 1000
lr_decayed_fn = (
  tf.keras.experimental.CosineDecayRestarts(
      initial_learning_rate,
      global_step,
      first_decay_steps))

You can pass this schedule directly into a tf.keras.optimizers.Optimizer as the learning rate. The learning rate schedule is also serializable and deserializable using tf.keras.optimizers.schedules.serialize and tf.keras.optimizers.schedules.deserialize.

Args:

  • initial_learning_rate: A scalar float32 or float64 Tensor or a Python number. The initial learning rate.
  • first_decay_steps: A scalar int32 or int64 Tensor or a Python number. Number of steps to decay over.
  • t_mul: A scalar float32 or float64 Tensor or a Python number. Used to derive the number of iterations in the i-th period
  • m_mul: A scalar float32 or float64 Tensor or a Python number. Used to derive the initial learning rate of the i-th period:
  • alpha: A scalar float32 or float64 Tensor or a Python number. Minimum learning rate value as a fraction of the initial_learning_rate.
  • name: String. Optional name of the operation. Defaults to 'SGDRDecay'.

Returns:

A 1-arg callable learning rate schedule that takes the current optimizer step and outputs the decayed learning rate, a scalar Tensor of the same type as initial_learning_rate.

Raises:

  • ValueError: if global_step is not supplied.

Methods

__call__

View source

__call__(step)

from_config

View source

from_config(
    cls,
    config
)

Instantiates a LearningRateSchedule from its config.

Args:

  • config: Output of get_config().

Returns:

A LearningRateSchedule instance.

get_config

View source

get_config()