@article{tidetokeninformeddepthexecutionforpertok, title = {TIDE: Token-Informed Depth Execution for Per-Token Early Exit in LLM Inference}, author = {Jaber Jaber and Osama Jaber}, year = {2026}, eprint = {2603.21365}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2603.21365}, }