@article{layercaketokenawarecontrastivedecodingwi, title = {LayerCake: Token-Aware Contrastive Decoding within Large Language Model Layers}, author = {Jingze Zhu and Yongliang Wu and Wenbo Zhu and Jiawang Cao and Yanqiang Zheng and Jiawei Chen and Xu Yang and Bernt Schiele and Jonas Fischer and Xinting Hu}, year = {2025}, eprint = {2507.04404}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2507.04404}, }