{"ID":5675895,"CreatedAt":"2026-07-03T01:40:09.565152011Z","UpdatedAt":"2026-07-04T16:50:11.910852832Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.01308","arxiv_id":"2607.01308","title":"Cache Merging as a Convergent Replicated State for Multi-Agent Latent Reasoning","abstract":"Multi-agent latent reasoning composes agents' KV-caches into one context for a final agent. Prior work (Agent Primitives) does this by concatenating caches along the sequence axis with RoPE re-encoding, which we call BagMerge. BagMerge is non-commutative, and the best input ordering is unpredictable, shifting with the regime, the latent-step budget, and the model scale. We make this exchange a convergent replicated state. First, CanonicalMerge fixes the layout by content: ordering caches by mean K-norm at a middle layer renders the merged cache byte-identical under any input permutation, verified algorithmically (arity N\u003c=5) and bit-for-bit on real Qwen3-1.7B and 4B state. Second, we separate the replicated state from decode-time layout: the state is a set of content-addressed latent fragments whose merge is set union, a state-based CvRDT (commutative, associative, idempotent, absorbing), and CanonicalMerge is its deterministic render. Because the render is byte-equivalent, every N=2 accuracy number carries over unchanged and re-delivered duplicates are absorbed rather than re-concatenated. On a partitioned-reasoning benchmark, CanonicalMerge matches the best BagMerge ordering in every regime-by-budget-by-ordering cell without knowing which order is best, trading a small, statistically insignificant accuracy margin for an unconditional structural guarantee. The behaviour transfers to real multi-document QA (HotpotQA), while the closest training-free output-fusion baseline (PackLLM) loses by 45 points at matched budget, placing cache-level merging in a regime distinct from output-level fusion. Finally, at k\u003e2 the approach transports and colocates latent traces but does not by itself compose them, which we characterize to motivate future work.","short_abstract":"Multi-agent latent reasoning composes agents' KV-caches into one context for a final agent. Prior work (Agent Primitives) does this by concatenating caches along the sequence axis with RoPE re-encoding, which we call BagMerge. BagMerge is non-commutative, and the best input ordering is unpredictable, shifting with the...","url_abs":"https://arxiv.org/abs/2607.01308","url_pdf":"https://arxiv.org/pdf/2607.01308v1","authors":"[\"Carlos Baquero\",\"Luís Brito\"]","published":"2026-07-01T17:05:04Z","proceeding":"cs.MA","tasks":"[\"cs.MA\"]","methods":"[\"Large Language Model\"]","has_code":false}