【CVPR '23】Learned Image Compression with Mixed Trasformer-CNN Architectures