AdaGrad

class vulkpy.nn.AdaGrad

Bases: Optimizer

AdaGrad Optimizer

Notes

This class implement AdaGrad [adagrad1].

References

[adagrad1]

Methods Summary

init_state(shape)

Initialize Optimizer state

Methods Documentation

init_state(shape: Iterable[int]) → AdaGradState

Initialize Optimizer state

Parameters:

shape (iterable of ints) – Shape of parameter

Returns:

Optimizer state

Return type:

AdaGradState

__init__(gpu: GPU, *, lr: float = 0.01, tau: float = 0.0, eps: float = 1e-08)

Initialize AdaGrad

Parameters:

• gpu (vulkpy.GPU) – GPU

• lr (float, optional) – AdaGrad parameter (learning rate). The default is 0.01.

• tau (float, optional) – AdaGrad parameter (initialial accumulator). The default is 0.

• eps (float, optional) – AdaGrad parameter (small positive). The default is 1e-8