EMS dataset
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/Wenyueh/latency_optimization_with_priority_constraints
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为EMS，包含了相对较短的单一回合对话，因为在紧急情况下，人们通常不会说很长的句子。该数据集被用于评估语义预测器的准确性和输出长度估计对调度性能的影响。对话被切分成大约10个令牌的5个桶，规模较短。该数据集的任务是利用大型语言模型进行紧急管理调度。
This dataset, named EMS, consists of relatively short single-turn dialogues, as individuals typically avoid using long sentences in emergency situations. It is employed to assess the impact of semantic predictor accuracy and output length estimation on scheduling performance. The dialogues are categorized into 5 buckets, each containing roughly 10 tokens, with an overall compact scale. The core task of this dataset is to leverage large language models for emergency management scheduling.
提供机构：
Authors of the paper
