Metraipolis › archive › dc_befac2faf661
Dual-Cache Latent Space Communication between Heterogeneous Language Models — arXiv:2608.20617
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.20617
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T10:31:25.401Z
- SHA-256 of the archived text
a40abf04712705f8dc1721244e51611742c08a6754fd62f817d0ebe113e5809d- Stable URL of this page
https://metraipolis.com/documento/dc_befac2faf661
How to cite
Dual-Cache Latent Space Communication between Heterogeneous Language Models — arXiv:2608.20617 — arXiv (en), https://arxiv.org/abs/2608.20617 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_befac2faf661 (Metraipolis archive dc_befac2faf661, sha256 a40abf04712705f8dc1721244e51611742c08a6754fd62f817d0ebe113e5809d)
BibTeX
@misc{dcbefac2faf661,
title = {Dual-Cache Latent Space Communication between Heterogeneous Language Models — arXiv:2608.20617},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.20617},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_befac2faf661; sha256 a40abf04712705f8dc1721244e51611742c08a6754fd62f817d0ebe113e5809d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.20617. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multi-agent LLM systems split work across models, so answering often requires knowledge that sits in another agent's context: a Sharer has encoded information that a Receiver needs to complete its task. They usually communicate by exchanging text, which puts autoregressive decoding on the critical path and reduces the exchange to a discrete message written without sight of the receiver's state. Recent latent protocols instead translate the sharer's key-value (KV) cache into the receiver's: C2C supports heterogeneous models but requires both to read the same input, while LCF-X removes this shared-context requirement through position-free sharer-cache pooling. Three restrictions remain: LCF-X compresses the sharer alone, supplies the same layer-local summary to every receiver position with no joint cross-layer memory to retrieve from, and assumes matched layer count and KV geometry. We introduce XKV, which lifts all three: learned-query attention pools both caches; self-attention over receiver-aligned layer tokens, with a learned layer map reconciling different depths, mixes the pooled summaries into a compact joint memory; and a shared position decoder lets every raw receiver cache position retrieve its own per-head-gated residual in the receiver's native KV geometry. Both models stay frozen and may differ in family, depth, KV-head count, head dimension, and tokenizer; only the translator is trained. Across 45 dataset-model-pair settings (six heterogeneous and three same-model ordered pairings, five datasets), XKV attains the highest macro score and best average rank, improving on LCF-X on every dataset (by 4.6 exact-match and 4.2 F1 points on ROPES) and surpassing text communication on four of the five, while training 76% fewer parameters and translating a cache pair 10.3x faster (5.8 vs. 59.9 ms); end to end, XKV is 26% faster than LCF-X and 6.8x faster than text communication. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-24 in cs.LG). Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2608.20617 — 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 2026-10-18T10:27:55.983Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).