Abstract: Embedding models are fundamental to modern Natural Language Processing (NLP), transforming textual data into dense vector representations that capture semantic meaning and relationships.

Since the generative adversarial network (GAN) was proposed by Ian Goodfellow et al. in 2014, it has been widely used in various fields. However, there are only a few works related to image ...