Abstract:Humans shift between different personas depending on social context. Large Language Models (LLMs) demonstrate a similar flexibility in adopting different personas and behaviors. Existing approaches, however, typically adapt such behavior through external knowledge such as prompting, retrieval-augmented generation (RAG), or fine-tuning. We ask: do LLMs really need external context or parameters to adapt to different behaviors, or do they already have such knowledge embedded in their parameters? In this work, we show that LLMs already contain persona-specialized subnetworks in their parameter space. Using small calibration datasets, we identify distinct activation signatures associated with different personas. Guided by these statistics, we develop a masking strategy that isolates lightweight persona subnetworks. Building on the findings, we further discuss: how can we discover opposing subnetwork from the model that lead to binary-opposing personas, such as introvert-extrovert? To further enhance separation in binary opposition scenarios, we introduce a contrastive pruning strategy that identifies parameters responsible for the statistical divergence between opposing personas. Our method is entirely training-free and relies solely on the language model's existing parameter space. Across diverse evaluation settings, the resulting subnetworks exhibit significantly stronger persona alignment than baselines that require external knowledge while being more efficient. Our findings suggest that diverse human-like behaviors are not merely induced in LLMs, but are already embedded in their parameter space, pointing toward a new perspective on controllable and interpretable personalization in large language models.
Кованова пояснила, что это происходит из-за ускоренной потери кальция, которую провоцируют фосфорная кислота и кофеин, присутствующие в коле. Помимо этого, диетолог напоминала, что сладкая газировка не рекомендуется людям с ожирением или диабетом.
С 28 февраля россияне смогут в течение нескольких дней наблюдать малый парад планет. О редком явлении рассказал доцент-практик департамента электроники, телекоммуникаций и приборостроения Политехнического института ДВФУ Петр Южаков в беседе с ТАСС.。业内人士推荐51吃瓜作为进阶阅读
The web browser has been at the core of the experience of using the internet basically since the internet was first created. But while they’ve always been key to the internet as a whole, until now, they’ve also been passive tools that you have to control. AI browsers are changing that in a pretty fundamental way. Instead of sitting there waiting for your input, these browsers can actually read pages, summarize content, dig through research, and take action on your behalf.
，推荐阅读WPS下载最新地址获取更多信息
Get editor selected deals texted right to your phone!
然而OpenAI先是成为了星门计划的核心成员，现在又被贴上了“军工复合体”的标签，技术民主化这个词已经离他们越来越远。，这一点在heLLoword翻译官方下载中也有详细论述