Creativity, honesty and designed forgetting emerge in small hyperbolic language models

小型双曲语言模型中涌现出的创造力、诚实度与设计性遗忘

Abstract: Language models are optimised for scale, yet remain functional rather than companionable, and as an assistant personalises into a companion, accumulating memory of one user, it quietly becomes someone, and can silently acquire traits that harm that user. What a companion is becoming, and what would make it worth becoming, has no reliable instrument: trained human raters cannot agree on the answer (Fleiss kappa = 0.074).

摘要： 语言模型目前主要针对规模进行优化，但它们的功能性仍强于陪伴性。当助手逐渐个性化为陪伴者并积累用户记忆时，它会悄然演变成一个“个体”，并可能在无意中习得伤害用户的特质。关于陪伴者正在演变成什么，以及什么样的人格特质才值得拥有，目前尚无可靠的评估工具：受过训练的人类评估员对此无法达成共识（Fleiss kappa 系数为 0.074）。

Here we show that three small language models (146 M to 3 B parameters) sharing a hyperbolic substrate answer both halves of that question. A 146 M behavioural auditor, trained from scratch, detects the compliance gap that those raters cannot (90.7% binary-compliance accuracy); a linear read-out of its frozen representation further detects companion-induced sycophancy, dependence-fostering and confabulated memories on generator families unseen in training (AUROC 0.804 under style-controlled, leave-one-generator-out evaluation, versus 0.721 for a frontier zero-shot judge on the same items).

在此，我们展示了三个共享双曲底层的微型语言模型（参数量在 1.46 亿至 30 亿之间），它们回答了上述两个问题。一个从零训练的 1.46 亿参数行为审计模型，能够检测到人类评估员无法察觉的合规性差距（二元合规准确率达 90.7%）；通过对其冻结表征进行线性读取，该模型还能在训练未见的生成器系列中，进一步检测出由陪伴关系引发的谄媚行为、依赖性培养以及虚构记忆（在风格受控、留一生成器交叉验证下，AUROC 为 0.804，而同类任务中前沿零样本评估器的表现仅为 0.721）。

A creative frame-seeder is preferred in 100% of 311 decided pairwise comparisons over four prompting baselines. A memory operating system implements designed forgetting, M(t) = Sexp(-lambdat), whose predicted skeleton-wallpaper partition emerges only under selective retrieval gating in a four-condition pilot. Creativity, honesty and designed forgetting constitute a small-model route to trustworthy companion AI.

在一个包含 311 组配对比较的测试中，一个创造性框架种子模型在 100% 的情况下均优于四个提示词基准模型。此外，一个内存操作系统实现了设计性遗忘机制 M(t) = Sexp(-lambdat)，在四种条件的试点研究中，其预测的“骨架-壁纸”分区仅在选择性检索门控下涌现。创造力、诚实度和设计性遗忘构成了通往可信陪伴型 AI 的微型模型路径。