{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "Your self-hosted model spends too long generating answers token by token.",
    "change": "Check whether a compatible draft model and serving integration exist before renting compute.",
    "baseline": "The same target with speculation off, then your existing drafter if you have one.",
    "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": "Accepted draft length, tokens per second, tail latency, output quality, and total GPU 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": "2610.26531",
  "title": "ReAnchor: Reusing Rejected Verifier States for Diffusion Drafting",
  "summary": "ReAnchor replaces the fixed single-block diffusion draft by reanchoring a pretrained drafter at shifted positions using verifier hidden states that standard speculative decoding discards after the first rejection, generating parallel draft blocks that extend coverage beyond the original block endpoint. It is training-free, so the cost is limited to implementation complexity and modest extra compute for the additional reanchored draft blocks, with no retraining or extra model weights. Gains can shrink or vanish at sampling temperatures above zero, depend on a compatible block diffusion drafter such as DFlash, and may not transfer to workloads or target models outside the evaluated set.",
  "rationale": "The method is training-free and works with a single-GPU-servable target plus a pretrained drafter, but its benefits are demonstrated mainly at greedy decoding and require an existing diffusion-drafter speculative decoding stack, so the reader should validate it on their own workload before production adoption.",
  "technique": "Reusing verifier hidden states for reanchored diffusion drafting",
  "family": "decodificacao_especulativa",
  "recommendation": "testar",
  "published": "2026-10-19",
  "as_of": "2026-10-20",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2610.26531",
  "independent_implementations": 1,
  "total_implementations": 2,
  "stars": 122,
  "citations": null,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "ReAnchor replaces the fixed single-block diffusion draft by reanchoring a pretrained drafter at shifted positions using verifier hidden states that standard speculative decoding discards after the first rejection, generating parallel draft blocks that extend coverage beyond the original block endpoint. It is training-free, so the cost is limited to implementation complexity and modest extra compute for the additional reanchored draft blocks, with no retraining or extra model weights. Gains can shrink or vanish at sampling temperatures above zero, depend on a compatible block diffusion drafter such as DFlash, and may not transfer to workloads or target models outside the evaluated set.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.26531",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "At T=0, largest gains on Gemma 4-12B: +19.4% acceptance length and +17.9% decoding speed; averaged over four target models and eight benchmarks with pretrained DFlash drafters, +15.2% acceptance length and +13.4% decoding speed.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.26531",
      "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": ""
}