tf.keras.activations.hard_sigmoid(x)
Defined in tensorflow/python/keras/activations.py.
Hard sigmoid activation function.
Faster to compute than sigmoid activation.
Arguments:
x: Input tensor.
Returns:
Hard sigmoid activation:
- 0 if x < -2.5
- 1 if x > 2.5
- 0.2 * x + 0.5 if -2.5 <= x <= 2.5.