Long-Context

DocPuzzle: A Process-Aware Benchmark for Evaluating Realistic Long-Context Reasoning Capabilities

论文概览论文标题：DocPuzzle: A Process-Aware Benchmark for Evaluating Realistic Long-Context Reasoning Capabilities 主要贡献：提出了一个专门评估大语言模型长文本推理能力的新基准数据规模：100个专家级问答问题，涵盖5个现实领域 ...

LongEval: A Comprehensive Analysis of Long-Text Generation Through a Plan-based Paradigm

论文概览论文标题：LongEval: A Comprehensive Analysis of Long-Text Generation Through a Plan-based Paradigm 研究机构：Multiple institutions 数据规模：166份跨三大领域的真实长文本样本核心创新：首创双范式长文本生成评估框架（直接生成 vs 规划生成） ...