Implement Swish Activation

ReLU has a hard corner at zero — its gradient jumps abruptly from 0 to 1. Swish smooths that out by weighting each input by its own sigmoid, giving a curve that dips slightly negative before rising, instead of clamping flat to zero:

\text{swish}(x) = x \cdot \sigma(x) = \frac{x}{1 + e^{-x}}

Task: write swish(values) applying this element-wise, rounded to 4 decimal places.

Unlike ReLU, Swish is not zero for every negative input — it dips slightly below zero before flattening out.
Zero maps to exactly zero, since σ(0) = 0.5 and 0 × 0.5 = 0.

Implement Swish Activation

Examples

Hints