Understanding Fine-tuning CLIP for Open-vocabulary Semantic Segmentation in Hyperbolic Space | Read Paper on Bytez