Metraipolis › archive › dc_657479df5cee

XBridge: Entity-Grounded Latent Bridge for Heterogeneous LLM Communication — arXiv:2608.11676

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.11676
License (as archived): CC0-1.0
Archived: 2026-10-19T05:15:27.383Z
SHA-256 of the archived text: e4b53060ccbdf2022dd47644cc15c7c0d00d2d97b601626ad8299b1de717b29b
Stable URL of this page: https://metraipolis.com/documento/dc_657479df5cee

How to cite

XBridge: Entity-Grounded Latent Bridge for Heterogeneous LLM Communication — arXiv:2608.11676 — arXiv (en), https://arxiv.org/abs/2608.11676 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_657479df5cee (Metraipolis archive dc_657479df5cee, sha256 e4b53060ccbdf2022dd47644cc15c7c0d00d2d97b601626ad8299b1de717b29b)

BibTeX

@misc{dc657479df5cee,
  title = {XBridge: Entity-Grounded Latent Bridge for Heterogeneous LLM Communication — arXiv:2608.11676},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.11676},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_657479df5cee; sha256 e4b53060ccbdf2022dd47644cc15c7c0d00d2d97b601626ad8299b1de717b29b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.11676. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Heterogeneous multi-agent LLM systems, where agents are powered by different model families, can outperform homogeneous configurations by reducing redundant reasoning patterns. Yet existing communication protocols either operate through text, discarding the sender's internal representations, or require architectural homogeneity for latent-level transfer. We identify the entity grounding problem in cross-architecture communication: cross-attention bridges that transfer continuous representations across different LLM families suffer from rare-token compression collapse, where entity identity is lost in the continuous bottleneck (bridge-only F1 ~30%). We propose XBRIDGE, a decode-free communication protocol that addresses this through two mechanisms. Lexical Anchor Mapping (LAM) maps the sender's original context tokens to the receiver's vocabulary, providing discrete entity anchors. A Latent Enrichment Bridge (LEB) lets the receiver query the sender's hidden states for contextual enrichment. The entity anchors ground the bridge's contextual signals to specific entities through the receiver's own self-attention. Across three model families (Llama, Qwen, and Mistral), seven benchmarks, and both communication directions, XBRIDGE outperforms text-based communication on all seven tasks for each model pair while achieving 11x lower latency, and in a same-architecture setting it also exceeds a KV-sharing baseline on six of seven tasks. LEB requires only 264M trainable parameters (3.8% of the receiver), is trained on a small balanced sample set, and adds negligible inference overhead.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-13 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.11676 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-19T05:08:46.638Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-13?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).