rztdl.dl.components.optimizers.primitive package

Submodules

rztdl.dl.components.optimizers.primitive.adam module

@created on: 12/15/19, @author: Himaprasoon, @version: v0.0.1

Description:

Sphinx Documentation Status:

class rztdl.dl.components.optimizers.primitive.adam.Adam(name: str, inputs: typing.Union[str, tensorflow.python.framework.ops.Tensor] = None, lr: float = 0.01, scopes: typing.Union[str, typing.List[str]] = None)[source]

Bases: tensorflow.python.keras.optimizer_v2.adam.Adam, rztdl.dl.components.optimizers.optimizer.Optimizer

Adam Optimizer

Parameters:
  • name (str) – Name of instance
  • inputs (Union[str, Tensor, None]) – Input Tensor
  • lr (float) – Learning rate
  • scopes (Union[str, List[str], None]) – list of Tags which can be used to train only specific layers during train
validate(inputs)[source]

rztdl.dl.components.optimizers.primitive.rms_prop module

@created on: 2020-01-31,
@author: shubham,
@version: v0.0.1

Description:

Sphinx Documentation Status: Complete

class rztdl.dl.components.optimizers.primitive.rms_prop.RMSprop(name: str = None, learning_rate: float = 0.001, rho: float = 0.9, momentum: float = 0.0, epsilon: float = 1e-07, centered: bool = False, inputs: typing.Union[str, tensorflow.python.framework.ops.Tensor] = None, outputs: str = None, scopes: typing.Union[str, typing.List[str]] = None)[source]

Bases: tensorflow.python.keras.optimizer_v2.rmsprop.RMSprop, rztdl.dl.components.optimizers.optimizer.Optimizer

Optimizer that implements the RMSprop algorithm

Parameters:
  • name (Optional[str]) – Name of the component
  • learning_rate (float) – learning rate for the optimizer
  • rho (float) – Discounting factor for the history/coming gradient
  • momentum (float) – A scalar tensor
  • epsilon (float) – Small value to avoid zero denominator.
  • centered (bool) – If True, gradients are normalized by the estimated variance of the gradient; if False, by the uncentered second moment
  • inputs (Union[str, Tensor, None]) – Input tensor/component name.
  • outputs (Optional[str]) – Output name.
  • scopes (Union[str, List[str], None]) – list of Tags which can be used to train only specific layers during train
parameter_validation(momentum)[source]
validate(inputs)[source]

rztdl.dl.components.optimizers.primitive.sgd module

@created on: 2020-01-31,
@author: shubham,
@version: v0.0.1

Description:

Sphinx Documentation Status: Complete

class rztdl.dl.components.optimizers.primitive.sgd.SGD(name: str, learning_rate: float = 0.01, momentum: float = 0.0, nesterov: bool = False, inputs: typing.Union[str, tensorflow.python.framework.ops.Tensor] = None, outputs: str = None, scopes: typing.Union[str, typing.List[str]] = None)[source]

Bases: tensorflow.python.keras.optimizer_v2.gradient_descent.SGD, rztdl.dl.components.optimizers.optimizer.Optimizer

Stochastic gradient descent and momentum optimizer

Parameters:
  • name (str) – Name of the component
  • learning_rate (float) – learning rate for the optimizer
  • momentum (float) – float hyperparameter >= 0 that accelerates SGD in the relevant direction and dampens oscillations
  • nesterov (bool) – Whether to apply Nesterov momentum.
  • inputs (Union[str, Tensor, None]) – Input tensor/component name.
  • outputs (Optional[str]) – Output name.
  • scopes (Union[str, List[str], None]) – list of Tags which can be used to train only specific layers during train
parameter_validation(momentum)[source]
validate(inputs)[source]

Module contents

@created on: 23/01/20,
@author: Umesh Kumar,
@version: v0.0.1

Description:

Sphinx Documentation Status: Complete