With no explanation, label text_A→text_B with either "strengthener" or "weakener".
text_A: A young boy giving a thumbs up to the camera. The boy is smiling at the camera.
text_B: The boy is half looking away from the camera with an ambiguous look on his face.
weakener.