tensornn.optimizers.RMSProp

class tensornn.optimizers.RMSProp(learning_rate: float = 0.001, decay: float = 0.9, epsilon: float = 1e-08)

Bases: Optimizer

Root Mean Square Propagation optimizer.

Methods

register	Register the optimizer with the model.
reset	Reset the optimizer.
set_lr	Set the learning rate of the optimizer.
source
step	Perform a step of optimization.

Attributes

model
learning_rate

__init__(learning_rate: float = 0.001, decay: float = 0.9, epsilon: float = 1e-08) → None

Initialize the optimizer.

Parameters:

• learning_rate – the learning rate of the optimizer

• decay – the decay rate of the optimizer

• epsilon – the epsilon value of the optimizer

__repr__()

Return repr(self).

register(model: TensorNNObject) → None

Register the optimizer with the model.

Parameters:

model – the model to register the optimizer with

reset() → None

Reset the optimizer.

set_lr(lr: float) → None

Set the learning rate of the optimizer.

Parameters:

lr – the learning rate to set

step() → None

Perform a step of optimization.