让大模型直接交换语义向量，提升协作效率
Direct Semantic Communication Between Large Language Models via Vector Translation
- 用可学习的映射在模型间传递语义向量，跳过原始文本
- 跨模型向量对齐度达0.538，30%混合强度即可有效引导生成
- 通用模型比指令微调模型更易传输语义，适合多智能体协作
在辩论、反思或工具调用等多智能体场景中，大语言模型（LLMs）通常以纯文本形式传递信息，导致大部分潜在语义丢失，并增加计算开销。本文通过向量翻译构建隐式桥梁，利用训练好的映射实现不同模型表示空间间的直接语义交互。在Llama-2-7B与Mistral-7B-Instruct之间训练的双编码器翻译器，平均余弦对齐度达到0.538。以30%的混合强度注入翻译后的向量，可在不破坏逻辑输出的前提下有效引导目标模型生成。双向评估显示2.01:1的传输不对称性，表明通用模型产生的表示更具可迁移性。该保守注入策略保持了计算稳定性，证明跨模型隐式通信可行，为共享意义而非仅传递文本的协作式AI系统提供了可能。
原文摘要 · Abstract (English)
In multi-agent settings, such as debate, reflection, or tool-calling, large language models (LLMs) pass messages as plain tokens, discarding most latent semantics. This constrains information transfer and adds unnecessary computational overhead. We form a latent bridge via vector translations, which use learned mappings that enable direct semantic exchange between representation spaces. A dual-encoder translator trained between Llama-2-7B and Mistral-7B-Instruct attains an average cosine alignment of 0.538. Injecting the translated vectors at 30 percent blending strength steers the target model's generation without destabilizing logits. Bidirectional evaluation shows a 2.01:1 transfer asymmetry, indicating that general-purpose models yield more transferable representations than instruction-tuned variants. This conservative injection preserves computational stability while demonstrating that cross-model latent communication is feasible, enabling collaborative AI systems that share meaning rather than tokens.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。