{"ID":2845115,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2511.03945","arxiv_id":"2511.03945","title":"Direct Semantic Communication Between Large Language Models via Vector Translation","abstract":"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.","short_abstract":"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 mapping...","url_abs":"https://arxiv.org/abs/2511.03945","url_pdf":"https://arxiv.org/pdf/2511.03945v1","authors":"[\"Fu-Chun Yang\",\"Jason Eshraghian\"]","published":"2025-11-06T00:43:29Z","proceeding":"cs.CL","tasks":"[\"cs.CL\",\"cs.AI\"]","methods":"[\"Large Language Model\",\"Language Model\"]","has_code":false}