These are the checkpoints of our proposed model from the paper 'Enhancing Cross-Modal Medical Image Segmentation through Compositionality', trained on the MMWHS and the CHAOS dataset in both directions.