generated_ideas

该数据集包含多个子集，分别来自不同顶级机器学习会议的oral论文（ICML 2024-2025、ICLR 2024-2026、NeurIPS 2024）。数据集共包含6个划分，总计8,617个样本，总体大小约41.4MB。每个样本包含6个字段：论文ID（字符串）、标题（字符串）、条件（字符串）、使用字段（字符串）、提供的上下文（字符串）和生成输出（字符串）。数据按会议年份和类型进行划分，每个划分的大小和样本数均有明确统计。

This dataset comprises multiple subsets sourced from oral papers published at top-tier machine learning conferences, including ICML 2024–2025, ICLR 2024–2026, and NeurIPS 2024. It contains 6 total splits, with 8,617 samples overall and a total size of approximately 41.4 MB. Each sample includes 6 fields: paper ID (string), title (string), condition (string), usage field (string), provided context (string), and generated output (string). The data is split by conference year and paper type, with explicit statistics available for the size and sample count of each split.