没有理解的趋同：当语言模型在表示上一致但在推理上分歧时
Convergence Without Understanding: When Language Models Agree on Representations but Disagree on Reasoning
AI总结：
本研究通过分析16个语言模型在800个推理问题上的表示相似性，发现模型在表示上趋同但推理策略不同，表明表示趋同反映的是共享的输入处理约束而非共享的推理策略。
摘要：
在不同目标和架构下训练的大型语言模型已被证明会发展出越来越相似的内部表示，这一观察被形式化为柏拉图式表示假说。这种表示趋同是否延伸到对共享表示进行操作的推理过程仍未得到检验。我们在800个涵盖数学、科学、常识和真实性的推理问题上，评估了来自8个家族（1.5B到72B参数）的16个语言模型的表示相似性，并按问题难度、计算阶段和因果相关性进行分层。我们的分析揭示了三种分离：难度反转，模型在它们共同失败的问题上趋同更多（中心核对齐[CKA] = 0.897），而在它们解决的问题上趋同较少（CKA = 0.830）；生成差距，决策前表示对齐（CKA = 0.875），而决策后表示分歧（CKA = 0.274）；以及附带正确性，共享信息可在模型间解码（66%的迁移准确率），但对预测的因果影响极小（在不同消融协议下翻转率为1.5%到5.5%）。这些结果表明，语言模型中的表示趋同反映了共享的输入处理约束而非共享的推理策略，对集成设计、可解释性迁移和模型相似性评估有直接影响。代码可在 https://github.com/Usama1002/convergence-without-understanding 获取。
英文摘要：
Large language models trained under diverse objectives and architectures have been shown to develop increasingly similar internal representations, an observation formalized as the Platonic Representation Hypothesis. Whether this representational convergence extends to the reasoning processes that operate over shared representations remains untested. We evaluate representational similarity across 16 language models from 8 families (1.5B to 72B parameters) on 800 reasoning problems spanning mathematics, science, commonsense, and truthfulness, stratifying by problem difficulty, computational stage, and causal relevance. Our analysis reveals three dissociations: a difficulty inversion, where models converge more on problems they collectively fail (Centered Kernel Alignment [CKA] = 0.897) than on those they solve (CKA = 0.830); a generation gap, where pre-decision representations align (CKA = 0.875) while post-decision representations diverge (CKA = 0.274); and epiphenomenal correctness, where shared information is decodable across models (66% transfer accuracy) but exerts minimal causal influence on predictions (1.5% to 5.5% flip rate across ablation protocols). These results indicate that representational convergence in language models reflects shared input processing constraints rather than shared reasoning strategies, with direct implications for ensemble design, interpretability transfer, and evaluations of model similarity. Code is available at https://github.com/Usama1002/convergence-without-understanding.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
