AI Mindmeld: A Latent Interconnect Protocol

Computer Science > Machine Learning arXiv:2610.30215 (cs) [Submitted on 21 Oct 2026] Title: AI Mindmeld: A Latent Interconnect Protocol Authors: Siavash Golkar, Jake Kovalic, Shirley Ho.

计算机科学 > 机器学习 arXiv:2610.30215 (cs) [提交于 2026 年 10 月 21 日] 标题：AI Mindmeld：一种潜在互联协议 作者：Siavash Golkar, Jake Kovalic, Shirley Ho。

Abstract: LLMs can act as scientific agents using tools to consult specialists and run experiments. Yet specialist information must enter their context, generally as text, images, audio or video: a bottleneck between models with rich, high-dimensional representations. We advocate direct representation exchange: a mindmeld conveying many properties, including information an LLM would not think to request.

摘要：大语言模型（LLM）可以作为科学智能体，利用工具咨询专家并运行实验。然而，专家信息必须以文本、图像、音频或视频等形式进入其上下文，这在具有丰富高维表征的模型之间形成了一个瓶颈。我们提倡直接的表征交换：一种能够传达多种属性的“思维融合”（mindmeld），包括大语言模型本身不会想到去请求的信息。

It should be extensible to new specialists, flexible across questions and tasks, and compositional across independently connected specialists without joint training. The Latent Interconnect Protocol (LIP) learns connectors mapping specialist representations into LLM internal states. Training only connectors keeps this practical as LLMs grow and evolve.

它应当能够扩展至新的专家，在不同问题和任务间保持灵活性，并能在无需联合训练的情况下，实现独立连接的专家之间的组合性。潜在互联协议（LIP）通过学习连接器，将专家表征映射到大语言模型的内部状态中。仅训练连接器使得该方法在大语言模型不断增长和演进的过程中依然实用。

We connect three protein and three small molecule specialists to Gemma 3 (4B and 12B parameters) and Kimi K2.6 (1T total, 32B active), freezing all pretrained models. Independently trained connectors provide useful information without changing the LLM or existing connectors, supporting familiar and new question types about unseen objects.

我们将三个蛋白质专家和三个小分子专家连接到 Gemma 3（4B 和 12B 参数）以及 Kimi K2.6（总计 1T，活跃参数 32B）模型上，并冻结所有预训练模型。独立训练的连接器在不改变大语言模型或现有连接器的情况下提供了有用的信息，支持了关于未知对象的熟悉及新型问题类型。

We demonstrate zero-shot transfer to cross-domain tasks without joint training. Task performance and compositionality generally improve with LLM capability, with especially strong gains in zero-shot task transfer. Limitations remain: current LLMs were not trained for specialist representations as native inputs; access does not ensure reliable interpretation, integration or use in reasoning.

我们展示了在无需联合训练的情况下，跨领域任务的零样本迁移能力。任务性能和组合性通常随大语言模型能力的提升而改善，在零样本任务迁移方面表现尤为突出。局限性依然存在：当前的大语言模型并未针对将专家表征作为原生输入进行训练；仅仅获得访问权限并不能确保可靠的解释、整合或在推理中的使用。

We call on specialist developers to provide consistent representation interfaces and, especially, on LLM developers to train models to receive, combine and reason with these inputs. These efforts could expand scientific agents’ information while preserving conventional tools’ flexibility.

我们呼吁专家开发者提供一致的表征接口，并特别呼吁大语言模型开发者训练模型以接收、组合这些输入并据此进行推理。这些努力可以在保持传统工具灵活性的同时，扩展科学智能体的信息获取能力。