@article{depthattentioncrosslayervaluemixingforla, title = {Depth-Attention: Cross-Layer Value Mixing for Language Models}, author = {Boyi Zeng and Yiqin Hao and Zitong Wang and Shixiang Song and He Li and Feichen Song and Yifan Liu and Ziwei He and Xinbing Wang and Zhouhan Lin}, year = {2026}, eprint = {2606.05014}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2606.05014}, }