JSSP Supervised Dataset

该数据集是一个监督学习数据集，包含大约12万个关于作业车间调度问题（JSSP）的自然语言描述及其可行解决方案，旨在用于训练大型语言模型（LLM）以执行调度任务。数据集涵盖了从2x2到20x20不同规模的问题，操作时长在5到500个单位之间。此外，还创建了一个大约1000个例子的独立评估数据集，用于评估模型性能。该数据集的规模约为12万个示例，任务针对的是作业车间调度问题（JSSP）。

This dataset is a supervised learning dataset containing approximately 120,000 natural language descriptions and their feasible solutions for Job Shop Scheduling Problems (JSSP), designed for training Large Language Models (LLMs) to perform scheduling tasks. The dataset covers problems of varying scales ranging from 2x2 to 20x20, with operation durations between 5 and 500 units. In addition, an independent evaluation dataset with approximately 1,000 examples was constructed to assess model performance. This dataset includes approximately 120,000 instances, with tasks focused on the Job Shop Scheduling Problem (JSSP).

搜集汇总

背景与挑战

背景概述

该数据集是一个包含12万条记录的作业车间调度问题(JSSP)数据集，具有5个关键字段，专门用于训练大型语言模型(LLMs)解决调度问题。数据集提供了不同作业和机器数量的组合，以及相应的自然语言格式输入输出对。