can I do multi-class classification having about 160 classes using sentence similarity. Actually, I have lot of data imbalance. My total dataset size is 1178. but, it has 160 classes. Some classes have high examples and some were too low even one or two. I am thinking to create embeddings for the examples belonging to each class. When the example is matching with the class embeddings then it will trigger the class. Is it a good approach in this scenario?