1 paper
Yuanhe Zhang, Xinyao Zhou, Haoran Gao +5
Uncontrolled repetition can prolong autoregressive generation in large language models (LLMs) and enable resource consumption attacks. Prior analyses of repetitive generation have…