{
  "schema_version": "1.1",
  "id": "archive:https://arxiv.org/abs/2610.37645v1",
  "slug": "2610-37645v1-12yb9up",
  "url": "https://feed7.dev/p/2610-37645v1-12yb9up",
  "title": "Multi-State Linear Attention with Generalized Second-Order Routing",
  "why_included": "MERLIN adds mean and variance estimates to how multi-state linear attention routes information from long-context segments. It improves reported RULER and LongBench results on two model variants, including against baselines using four times more memory.",
  "summary": "MERLIN tracks the first and second moments of each segment’s keys, letting a query estimate both mean and variance in its alignment with that segment. On **Gated DeltaNet-1.3B** and **Mamba-2-780M**, it improves reported RULER and LongBench results, including over some baselines using **4× more memory**.",
  "practical_implication": "Builders exploring long-context models can watch this approach as a way to improve how compressed segment memories are read. It suggests routing quality may matter alongside the amount of memory available.",
  "agent_context": "MERLIN tracks the first and second moments of each segment’s keys, letting a query estimate both mean and variance in its alignment with that segment. On **Gated DeltaNet-1.3B** and **Mamba-2-780M**, it improves reported RULER and LongBench results, including over some baselines using **4× more memory**.\n\nBuilders exploring long-context models can watch this approach as a way to improve how compressed segment memories are read. It suggests routing quality may matter alongside the amount of memory available.\n\nThese are results on multi-state variants and specific long-context evaluations. The material does not say how the method behaves in deployed agent workloads or across other model families.",
  "source": {
    "name": "arXiv",
    "url": "https://arxiv.org/abs/2610.37645v1",
    "published_at": "2026-10-27T17:55:49.000Z"
  },
  "source_class": "blog_post",
  "content_type": "Paper",
  "layer": "model",
  "domains": [
    "research"
  ],
  "topics": [
    "open-models",
    "context-engineering",
    "retrieval"
  ],
  "verification": {
    "status": "needs_review",
    "label": "Needs Review",
    "method": "unverified",
    "verified_at": null
  },
  "uncertainty": [
    "These are results on multi-state variants and specific long-context evaluations. The material does not say how the method behaves in deployed agent workloads or across other model families."
  ],
  "connected_context": {
    "meaning": "MERLIN adds first- and second-moment routing to compressed segment memories and reports gains on two long-context model families. It suggests that how a model reads stored segments can matter alongside memory capacity, while leaving performance in deployed agent workloads unestablished.",
    "corpus_size": 944,
    "generated_at": "2026-10-28T11:08:13.296Z",
    "connections": [
      {
        "title": "ConvMem: Convolutional Memory for Long-Context Reasoning",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2609.10441v1",
        "feed7_url": "https://feed7.dev/p/2609-10441v1-066ygq8",
        "reason": "Both change how long inputs are represented and accessed: ConvMem builds hierarchical summaries, while MERLIN routes queries using segment key moments. Comparing them could clarify the tradeoff between memory organization and readout quality; neither supplied summary establishes coding-agent value."
      },
      {
        "title": "Structured Memory for Edge Language Models: Persistent Context and Corpus Retrieval via O(1) SSM State Injection",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2608.02560v1",
        "feed7_url": "https://feed7.dev/p/2608-02560v1-1u8ls14",
        "reason": "PRECOG injects reusable corpus states into compatible SSMs to reduce repeated prefill, while MERLIN targets query routing over compressed segment memories and includes Mamba-2 results. They address distinct parts of SSM context handling, and neither summary establishes RAG-equivalent quality or deployed-agent performance."
      }
    ]
  },
  "lifecycle": "Current",
  "published_at": "2026-10-27T17:55:49.000Z",
  "modified_at": "2026-10-27T17:55:49.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/2610-37645v1-12yb9up",
    "json": "https://feed7.dev/p/2610-37645v1-12yb9up.json",
    "markdown": "https://feed7.dev/p/2610-37645v1-12yb9up.md"
  }
}