Update README.md

All the segmentation models trained on ADE20K registred the index 0 to the wall (not the background). By adding [0, 0, 0] at the index 0 everything will be shifted to the right by one step and leads to incorrect results.

README.md

@@ -79,7 +79,6 @@ $ pip install diffusers transformers accelerate
 import numpy as np
 
 ada_palette = np.asarray([
-      [0, 0, 0],
       [120, 120, 120],
       [180, 120, 120],
       [6, 230, 230],