arXiv cs.CL· Gunhee Cho, Yun-Gyung Cheong·· 13 小时前AI 评分31
Personality Latents：用本人写作让冻结语言模型模仿个人性格
Personality Latents: Conditioning Frozen Language Models on Individuals from Their Own Writing
AI 导读
Personality Latents（PL）无需性格标签或自撰档案，直接从个人写作中学习隐变量，并以软前缀注入冻结语言模型。在 MBTI 各轴上，PL 的 AUC 超过 persona embedding 与基于提示词的基线，且在三种嵌入空间评分下排名一致。用 Qwen2.5-3B-Instruct 学到的隐变量仅通过新适配器即可迁移到 Llama-3.1-8B-Instruct。
正文 · 原文
Abstract:Controlling the personality of a language model usually relies on written profiles or named trait categories, so it can only express what an author writes down or what a taxonomy names. We instead propose Personality Latents (PL), which learn a latent for each person from their own writing, with no personality labels and no authored profile, and inject it into a frozen language model as a soft prefix. Following generative latent optimization, each training person receives a free latent that is optimized through the frozen model itself, while an encoder learns to predict these latents so that a new person needs only one forward pass, with optional refinement at test time. We evaluate on held-out people by having the conditioned model answer neutral and interview questions and reading personality from its answers with scorers trained on real human writing. On the MBTI axes PL reaches an AUC of , ahead of persona embedding and prompt-based baselines under identical settings, and the ranking holds when the answers are scored in three different embedding spaces. A latent learned with Qwen2.5-3B-Instruct also transfers personality to Llama-3.1-8B-Instruct through a new adapter alone.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.36805 [cs.CL] |
| (or arXiv:2610.36805v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.36805 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Gunhee Cho [view email]
[v1]
Tue, 27 Oct 2026 08:08:06 UTC (605 KB)
来源：arXiv cs.CL · arxiv.org