Module Disctinction

A submodule implementing functions to get class of abstract torch.nn.Module.

Examples

>>> import torch.nn as nn
>>> from monitorch.lens.module_distinction import isactivation, isconv
>>> isactivation(nn.ReLU())
True
>>> isactivation(nn.Dropout())
False
>>> isconv(nn.BatchNorm1d(10))
False
>>> isconv(nn.Conv2d(1, 1, 1))
True

monitorch.lens.module_distinction.isactivation(module: Module) → bool

Checks if provided module is an activation function module. Returns False for torch.nn.MultiheadAttention.

Parameters:

module (torch.nn.Module) – Module to be checked

monitorch.lens.module_distinction.isconv(module: Module) → bool

Checks if provided module is a convolution module.

Parameters:

module (torch.nn.Module) – Module to be checked

monitorch.lens.module_distinction.isdropout(module: Module) → bool

Checks if provided module is a dropout module.

Parameters:

module (torch.nn.Module) – Module to be checked

monitorch.lens.module_distinction.islinear(module: Module) → bool

Checks if provided module is a linear non-convolution module.

Parameters:

module (torch.nn.Module) – Module to be checked

monitorch.lens.module_distinction.isnormalization(module: Module) → bool

Checks if provided module is a normalization module.

Parameters:

module (torch.nn.Module) – Module to be checked