Graph Convolutions Enrich the Self-Attention in Transformers! | Read Paper on Bytez