With no explanation, label text_A→text_B with either "strengthener" or "weakener".
text_A: Brown-haired woman walking and staring at camera. Woman with sunglasses on.
text_B: The brown-haired woman has an intense stare at the camera.
weakener.