Skip to content

Step — Hard Threshold Activation

Maps values above threshold (default 0.0) to 1, others to 0. Mainly useful for explicit thresholding or post-processing.


Overview

Not smooth; provides limited gradient information. Useful when a hard decision boundary is required rather than a differentiable transition.


Key Features

  • Hard binary output (0 / 1)
  • Configurable threshold
  • Limited gradient for training

Parameters

Parameter Type Default Description
threshold float 0.0 Decision boundary
name str \| None None Layer name
verbose bool False Debug logging

Methods

forward(x, training=True)

Return the activation applied to input.

  • x: Tensor or array-like — Input values of any shape
  • Returns: Tensor with same shape as x

Example Usage

import numpy as np
from machinegnostics.magnet import Step

activation = Step()
result = activation(np.array([[0.25, 0.75]]))
print(result.data.tolist())  # [[0.0, 1.0]]

Author: Nirmal Parmar