generalized zero-shot learning (GZSL) has emerged as a pivotal research
domain in computer vision, owing to its capability to recognize objects that
have not been seen during training. Despite the significant progress achieved
by generative techniques in converting traditional GZSL to