Streamliners Dataset

该数据集是一个基准集合，包含了10个为评估大型语言模型生成的简化器而生成的约束满足问题。这些问题的实例包括平衡不完全区块设计（BIBD）、汽车序列（CS）、社交高尔夫球手（SG）、船舶装载（VL）、黑洞（BH）、地毯裁剪（CC）、超图着色（HC）、披萨配料划分（PIP），以及CS和HC问题的混淆版本。该数据集规模涵盖10个问题，每个问题包含15个训练实例和至少50个测试实例。任务内容为解决约束满足问题。

This dataset is a benchmark collection consisting of 10 constraint satisfaction problems (CSPs) generated to evaluate simplifiers developed for large language models (LLMs). The problem instances include Balanced Incomplete Block Design (BIBD), Car Sequencing (CS), Social Golfers (SG), Vehicle Loading (VL), Black Hole (BH), Carpet Cutting (CC), Hypergraph Coloring (HC), Pizza Topping Partition (PIP), as well as the obfuscated variants of the CS and HC problems. This dataset covers 10 problem types, each of which contains 15 training instances and at least 50 test instances. The core task is to solve constraint satisfaction problems.