量化会重构大模型的元认知结构，而非简单降低性能。
Quantisation Reshapes the Metacognitive Geometry of Language Models
- 量化改变模型在各领域的元认知监控效率分布。
- 艺术文学领域监控能力从差变优，地理领域则相反，但判断准确性不变。
- 依赖元认知比率的系统需警惕量化影响，用AUROC_2更可靠。
我们发现，模型量化并未均匀降低大语言模型的领域级元认知效率，而是重构了其结构。在相同3,000个问题上对Llama-3-8B-Instruct进行Q5_K_M与f16精度评估，四个知识领域的M-ratio在不同格式间无相关性（Spearman rho = 0.00）。艺术与文学领域从最低监控（M-ratio = 0.606，Q5_K_M）变为最高监控（1.542，f16），地理领域则从良好监控（1.210）变为欠监控（0.798）。然而，Type-2 AUROC在两种格式下完全稳定（rho = 1.00），说明重构仅源于M-ratio归一化，而非底层判别信号变化。该发现源自一项预注册的元认知改进实验：针对诊断出的弱领域施加置信度放大SFT，设置预算匹配的无关与错误处方对照组。四条验证假设均不显著（10,000次自助采样，种子=42）。训练成功重塑置信度分布，使科学领域NLP差距从0.076翻倍至0.152，但未提升meta-d'，因诊断表现无法跨格式迁移。任何依赖领域级M-ratio的系统都存在未被审视的推理格式依赖。使用AUROC_2的系统更安全。代码、预注册和试验级数据均已公开。
原文摘要 · Abstract (English)
We report that model quantisation restructures domain-level metacognitive efficiency in LLMs rather than degrading it uniformly. Evaluating Llama-3-8B-Instruct on the same 3,000 questions at Q5_K_M and f16 precision, we find that M-ratio profiles across four knowledge domains are uncorrelated between formats (Spearman rho = 0.00). Arts & Literature moves from worst-monitored (M-ratio = 0.606 at Q5_K_M) to best-monitored (1.542 at f16). Geography moves from well-monitored (1.210) to under-monitored (0.798). However, Type-2 AUROC profiles are perfectly stable across formats (rho = 1.00), localising the restructuring to the M-ratio normalisation rather than the underlying discrimination signal. This finding emerged from a pre-registered attempt to improve metacognition through domain-conditional training. We prescribed confidence-amplification SFT for the diagnosed weak domain, with matched-budget agnostic and wrong-prescription controls. All four confirmatory hypotheses were null (10,000 bootstrap resamples, seed = 42). The training successfully reshaped confidence distributions, doubling the NLP gap in Science from 0.076 to 0.152, but did not improve meta-d' because the diagnostic profile did not transfer across formats. Any system relying on domain-level M-ratio profiles has an unexamined dependency on inference format. Systems using AUROC_2 are safer. We release all code, pre-registrations, and trial-level data.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。