eliteresearch/researchscope-papers
收藏资源简介:
ResearchScope Papers是一个由ResearchScope维护的开放计算机科学研究论文数据集,通过GitHub Actions自动更新。数据集包含102,058篇论文的原始元数据、473,434条指令调优行,以及A*论文的逐章节微调行。数据来源包括arXiv、OpenAlex、ACL Anthology、OpenReview、PMLR、CVF和Semantic Scholar等,涵盖NeurIPS、ICML、ICLR、ACL、EMNLP、CVPR、AAAI、IJCAI、JMLR、TMLR、TACL、TPAMI、NMI等多个顶级会议和期刊。数据集文件包括:papers.jsonl(原始论文元数据,如标题、摘要、作者、会议、年份、标签和分数)、instruct.jsonl(指令调优对,用于总结、关键贡献、重要性解释和通俗英语转换)和sections.jsonl(A*论文的逐章节微调行,包含摘要、引言、相关工作、方法、实验、结果和结论等真实正文文本,可用于训练逐章节写作代理)。
ResearchScope Papers is an open CS research paper dataset maintained by ResearchScope, updated automatically via GitHub Actions. It includes 102,058 papers (raw metadata), 473,434 instruction-tuning rows, and per-section fine-tuning rows for A* papers. Sources include arXiv, OpenAlex, ACL Anthology, OpenReview, PMLR, CVF, and Semantic Scholar, with venues such as NeurIPS, ICML, ICLR, ACL, EMNLP, CVPR, AAAI, IJCAI, JMLR, TMLR, TACL, TPAMI, and NMI. The dataset files are: papers.jsonl (raw paper metadata — title, abstract, authors, venue, year, tags, scores), instruct.jsonl (instruction-tuning pairs — summarize, key contribution, why it matters, plain English), and sections.jsonl (per-section fine-tuning rows for A* papers — real body text of abstract, introduction, related_work, method, experiments, results, conclusion, filterable by the section field to train a per-section writing agent).





