keraflow.optimizers.Adamax Class Reference

Adamax optimizer from Adam paper's Section 7. More...

Inheritance diagram for keraflow.optimizers.Adamax:

Public Member Functions
def	__init__
	Default parameters follow those provided in the paper. More...
Public Member Functions inherited from keraflow.optimizers.Optimizer
def	__init__

Detailed Description

Adamax optimizer from Adam paper's Section 7.

It is a variant of Adam based on the infinity norm.

• Reference: Adam - A Method for Stochastic Optimization

Constructor & Destructor Documentation

def keraflow.optimizers.Adamax.__init__	(		self,
			lr = 0.002,
			beta_1 = 0.9,
			beta_2 = 0.999,
			epsilon = 1e-8,
			kwargs
	)

Default parameters follow those provided in the paper.

Parameters

lr	float >= 0. Learning rate.
beta_1/beta_2	floats, 0 < beta < 1. Generally close to 1.
epsilon	float >= 0. Fuzz factor.
kwargs	see Optimizer.__init__