DeepReview

DeepReviewer-13K is a curated dataset designed to facilitate the training of Large Language Models (LLMs) for academic paper review. It captures a human-like deep thinking process in peer review, featuring structured annotations and intermediate reasoning steps. This dataset addresses the scarcity of high-quality, structured review data and provides a valuable resource for advancing automated scientific evaluation.

DeepReviewer-13K是一款精心构建的精选数据集，旨在助力面向学术论文审稿任务的大语言模型（Large Language Model，LLM）训练。该数据集捕捉了同行评审过程中类人的深度思考流程，涵盖结构化标注与中间推理步骤。本数据集解决了高质量结构化审稿数据稀缺的痛点，为推动自动化科学评估技术的发展提供了宝贵的研究资源。