why to pad features with zero vector？

[zhanking321]

Excuse me, I wonder why to pad features with zero vector？

---

if not features is None:
    # pad with dummy zero vector
    features = np.vstack([features, np.zeros((features.shape[1],))])

---