{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "Your workload might fit a smaller or differently structured model.",
    "change": "Check for available weights and a compatible runtime before planning a product trial.",
    "baseline": "Your current model on identical held-out tasks and the same resource budget.",
    "steps": [
      "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.",
      "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.",
      "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.",
      "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test."
    ],
    "measure": "Task success, latency, memory, context limits, and migration cost.",
    "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation."
  },
  "builder_review": null,
  "arxiv_id": "2610.11214",
  "title": "Bridging KV-Cache Quantization and Linear Attention: From Theory to Pretrained Weight Migration",
  "summary": "RAM-Net replaces full softmax attention with a recurrent fixed-size slot state addressed by soft assignments over a discrete codebook, unifying KV-cache quantization and linear attention, so inference memory becomes constant in sequence length rather than growing with the KV cache. The cost is a per-model weight-migration training run of roughly 500M tokens, plus the engineering complexity of a nonstandard attention implementation, and a measured quality gap of about 13% of the teacher's task accuracy. It can fail on workloads where the slot state introduces interference between aggregated keys, on tasks outside the evaluated commonsense and knowledge benchmarks (notably long-range retrieval), and where no migration compute budget exists.",
  "rationale": "The constant-memory recurrent state is attractive for a single 24 GB GPU serving long contexts, but the required 500M-token migration per model and the unquantified memory savings demand small-scale validation before production adoption.",
  "technique": "RAM-Net: soft-quantized linear attention via weight migration",
  "family": "arquitetura_eficiente",
  "recommendation": "testar",
  "published": "2026-10-08",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2610.11214",
  "independent_implementations": 0,
  "total_implementations": 1,
  "stars": 4,
  "citations": 0,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "RAM-Net replaces full softmax attention with a recurrent fixed-size slot state addressed by soft assignments over a discrete codebook, unifying KV-cache quantization and linear attention, so inference memory becomes constant in sequence length rather than growing with the KV cache. The cost is a per-model weight-migration training run of roughly 500M tokens, plus the engineering complexity of a nonstandard attention implementation, and a measured quality gap of about 13% of the teacher's task accuracy. It can fail on workloads where the slot state introduces interference between aggregated keys, on tasks outside the evaluated commonsense and knowledge benchmarks (notably long-range retrieval), and where no migration compute budget exists.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.11214",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "Recovers an average of 87.1% of pretrained teachers' accuracy gains over random guessing across six commonsense and knowledge tasks, using a 500M-token migration budget per model across nine models from 0.3B to 7B.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.11214",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "exposure_map": [
    {
      "dimension": "quality",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compute",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "latency",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "operations",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compatibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "security",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "data_and_training",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "reproducibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "risks": [],
  "minimum_test": [],
  "open_questions": [],
  "independent_tests": [],
  "lab_reproductions": [],
  "report_available": false,
  "validation_tier": "unknown",
  "evidence_tier": "unknown",
  "infrastructure_basis": "unknown",
  "equations": [],
  "equations_status": "not_fetched",
  "core_kind": "",
  "equations_fetched_at": "",
  "equations_source_url": "",
  "equations_source_sha256": "",
  "exposures_reviewed_at": "",
  "exposures_source_sha256": ""
}