arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2501.14082cs.CLcs.AIcs.LG
收藏

语言模型智能体之间的激活通信

Communicating Activations Between Language Model Agents

Kempner Institute for AI, Harvard University(哈佛大学肯普纳人工智能研究所)

Vignav Ramesh, Kenneth Li

AI总结:

本文提出让语言模型智能体通过中间层激活而非自然语言进行通信,在零额外参数和数据下提升推理性能,并显著降低计算成本。

摘要:

多个语言模型(LM)智能体之间的通信已被证明能够扩展语言模型的推理能力。虽然自然语言一直是语言模型间通信的主要媒介,但这一点是否应作为标准并不显然:自然语言通信不仅会产生高昂的推理成本,且该成本会随着智能体数量和消息数量的增加而迅速上升,而且解码过程会过度抽象掉大量本可以从内部激活中获取的丰富信息。在这项工作中,我们提出一种简单的技术,使语言模型通过激活进行通信;具体而言,我们在中间层暂停语言模型 $\ extit{B}$ 的计算,将其当前激活与另一个语言模型 $\ extit{A}$ 的中间激活通过某个函数 $\ extit{f}$ 进行组合,然后将 $\ extit{f}$ 的输出传入 $\ extit{B}$ 的下一层,并继续前向传播直到解码完成。该方法在零额外参数和零额外数据的情况下将语言模型扩展到新任务上,并且相比自然语言通信节省了大量计算。我们在两种实验设置——多玩家协调博弈和推理基准——上使用多种函数形式 $\ extit{f}$ 测试了我们的方法,发现它在各数据集上相比自然语言通信实现了最高 $27.0\%$ 的提升,而计算量不到其 $1/4$,这说明了激活作为语言模型之间通信的替代“语言”具有优越性和鲁棒性。

英文摘要:

Communication between multiple language model (LM) agents has been shown to scale up the reasoning ability of LMs. While natural language has been the dominant medium for inter-LM communication, it is not obvious this should be the standard: not only does natural language communication incur high inference costs that scale quickly with the number of both agents and messages, but also the decoding process abstracts away too much rich information that could be otherwise accessed from the internal activations. In this work, we propose a simple technique whereby LMs communicate via activations; concretely, we pause an LM $\textit{B}$'s computation at an intermediate layer, combine its current activation with another LM $\textit{A}$'s intermediate activation via some function $\textit{f}$, then pass $\textit{f}$'s output into the next layer of $\textit{B}$ and continue the forward pass till decoding is complete. This approach scales up LMs on new tasks with zero additional parameters and data, and saves a substantial amount of compute over natural language communication. We test our method with various functional forms $\textit{f}$ on two experimental setups--multi-player coordination games and reasoning benchmarks--and find that it achieves up to $27.0\%$ improvement over natural language communication across datasets with $<$$1/4$ the compute, illustrating the superiority and robustness of activations as an alternative "language" for communication between LMs.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑