LG-Bench：生命科学领域首个图结构评测基准
LG-Bench: A Graph-Structured Evaluation Benchmark for Life Sciences
arXiv 上的新论文，把 1 万多道医学生化题连成语义图来测大模型，GCS 和 KBS 两个指标能看出模型哪里概念不一致，做医学方向评测的可以看看。
LG-Bench 是首个图结构的生命科学评测基准，包含超过 10,000 道覆盖医学、生物学、化学的选择题。该基准通过双向匹配与语义相似度算法构建加权评测图，节点为题目，边权重反映题目间的语义关系。在此图拓扑上设计了两个新指标：Global Coherence Score（GCS）衡量模型在语义相关邻域内的回答一致性，Knowledge Balance Score（KBS）分析模型错误在图上的分布以定位概念盲区。这一框架把评测从平面准确率转向结构感知分析，用于细粒度比较不同 LLM 的概念连贯性与知识组织模式。
LG-Bench: A Graph-Structured Evaluation Benchmark for Life Sciences
Traditional evaluation benchmarks reduce inherently interconnected scientific knowledge in life sciences into flat lists of questions, disregarding the underlying topological structure of the knowledge. We introduce, the first graph-structured benchmark for life sciences, featuring over 10,000 high-quality multiple-choice questions across medicine, biology, and chemistry. Our approach constructs a weighted evaluation graph using bidirectional matching and semantic similarity algorithms, where nodes represent questions and edge weights capture their semantic relationships. Leveraging this graph topology, we design two novel evaluation metrics. The Global Coherence Score (GCS) measures a model's consistency within semantically related neighborhoods, while Knowledge Balance Score (KBS) analyzes how model errors are distributed across the graph to reveal conceptual blind spots. LG-Bench facilitates fine-grained comparison of LLMs by surfacing differences in conceptual coherence and patterns of knowledge organization across models. Our framework shifts the evaluation paradigm from flat accuracy metrics to structure-aware analysis, offering a new lens for diagnosing and improving LLM performance in the life sciences domain.