AxiomicLabs/LogicMark
收藏资源简介:
LogicMark是一个程序生成的基准测试,用于评估语言模型中的符号逻辑推理能力。每个问题呈现一组变量相等/不相等的前提,并要求模型识别哪个结论必然成立。与基于知识的基准不同,LogicMark不包含模型可能从预训练中记忆的任何事实。每个问题都使用抽象变量名(a, b, c, ...)新鲜生成,因此模型无法通过模式匹配训练数据,必须实际进行推理。这使得LogicMark成为对内在推理能力的直接探测:即通过训练构建到模型权重中的逻辑结构,独立于世界知识或表面启发式方法。评估采用对数似然多项选择方式,无需思维链或提示技巧,模型仅根据为正确完成分配概率的能力进行评分。
LogicMark is a procedurally generated benchmark for evaluating symbolic logic in language models. Each problem presents a set of variable equality/inequality premises and asks the model to identify which conclusion necessarily follows. Unlike knowledge-based benchmarks, LogicMark contains no facts a model could have memorised from pretraining. Every problem is generated fresh from abstract variable names (a, b, c, ...), so a model cannot pattern-match to training data - it must actually reason. This makes LogicMark a direct probe of intrinsic reasoning capability: the logical structure that has been built into the models weights through training, independent of world knowledge or surface-level heuristics. Evaluation is log-likelihood multiple-choice — no chain-of-thought, no prompting tricks. Models are scored purely on how well they assign probability to the correct completion.




