@misc{meng2025transmla, title={TransMLA: Multi-head Latent Attention Is All You Need}, author={Fanxu Meng and Zengwei Yao and Muhan Zhang}, year={2025}, eprint={2502.07864}, archivePrefix={arXiv}, primaryClass={cs.LG}, url={https://arxiv.org/abs/2502.07864} }