{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "A self-hosted model uses more compute or memory than your workload can justify.",
    "change": "Check for a usable pruned checkpoint and runtime support before attempting model surgery.",
    "baseline": "The original unpruned model at the same precision, context, and load.",
    "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 quality, actual GPU memory, serving throughput, and preparation 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": "2610.11332",
  "title": "ReCal: Calibrating Structured Pruning for On-Policy Distillation Recovery",
  "summary": "ReCal modifies the calibration step of existing structured pruning methods by reweighting calibration statistics using forward KL between the unpruned teacher and a pruned probe, so that pruning criteria preserve teacher-supported predictions; it does not replace pruning itself or the on-policy distillation recovery loop. Cost is additional forward passes with the teacher and probe over calibration data before pruning, which is modest on a single 24 GB GPU and adds no serving latency since it is an offline preprocessing step. It can fail when the pruning damage it targets does not align with tokens that matter for downstream OPD trajectories, when the team lacks an OPD pipeline to realize the recovery gains, and when gains do not transfer to domains beyond the evaluated math and code reasoning tasks.",
  "rationale": "The method is a plug-and-play addition to existing pruning criteria with low compute overhead, but its value depends on the team already doing structured pruning plus on-policy distillation, so workload-specific validation is warranted before adoption.",
  "technique": "Recovery-aware calibration before structured pruning for OPD",
  "family": "esparsidade_e_poda",
  "recommendation": "testar",
  "published": "2026-10-08",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2610.11332",
  "independent_implementations": 0,
  "total_implementations": 1,
  "stars": 122,
  "citations": null,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "ReCal modifies the calibration step of existing structured pruning methods by reweighting calibration statistics using forward KL between the unpruned teacher and a pruned probe, so that pruning criteria preserve teacher-supported predictions; it does not replace pruning itself or the on-policy distillation recovery loop. Cost is additional forward passes with the teacher and probe over calibration data before pruning, which is modest on a single 24 GB GPU and adds no serving latency since it is an offline preprocessing step. It can fail when the pruning damage it targets does not align with tokens that matter for downstream OPD trajectories, when the team lacks an OPD pipeline to realize the recovery gains, and when gains do not transfer to domains beyond the evaluated math and code reasoning tasks.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.11332",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "Up to 16.7 percentage points improvement on AIME after on-policy distillation recovery, plus gains in most code-generation comparisons, across multiple models and pruning methods.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.11332",
      "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": ""
}