Abstract: Transformers have demonstrated outstanding performance in learned image compression (LIC) due to their high capacity for modeling complex dependencies. However, existing methods employ ...