{
  "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": "2610.00367",
  "title": "MoRA: MoE Pruning via Router Bias Learning and Expert Approximation",
  "summary": "Replaces heuristic expert-ranking criteria and expensive expert-subset search with a learnable per-expert router bias optimized on language-modeling loss plus a routing-diversity regularizer, followed by affine approximation of pruned experts from retained ones. Costs a calibration-style optimization pass over the router biases and approximation parameters, adds pipeline complexity, and pruned models will still show some quality degradation that grows with the pruning ratio. Can fail when removed experts carried rare but important routing behaviors that the affine approximation cannot reconstruct, and results are validated only on three specific MoE architectures, so transfer to other models is unverified.",
  "rationale": "Expert pruning directly addresses the reader's memory constraint by shrinking MoE checkpoints toward a 24 GB GPU budget, but the claimed gains are benchmark-relative rather than absolute and the procedure requires a per-model optimization pass, so it warrants a small-scale validation on the reader's exact model and workload before production adoption.",
  "technique": "MoE expert pruning via learned router bias",
  "family": "esparsidade_e_poda",
  "recommendation": "testar",
  "published": "2026-09-30",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2610.00367",
  "independent_implementations": 0,
  "total_implementations": 1,
  "stars": 4,
  "citations": null,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "Replaces heuristic expert-ranking criteria and expensive expert-subset search with a learnable per-expert router bias optimized on language-modeling loss plus a routing-diversity regularizer, followed by affine approximation of pruned experts from retained ones. Costs a calibration-style optimization pass over the router biases and approximation parameters, adds pipeline complexity, and pruned models will still show some quality degradation that grows with the pruning ratio. Can fail when removed experts carried rare but important routing behaviors that the affine approximation cannot reconstruct, and results are validated only on three specific MoE architectures, so transfer to other models is unverified.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.00367",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "Removes 25% and 50% of routed experts per MoE layer on Qwen3-30B-A3B, DeepSeek-V2-Lite, and Moonlight-16B-A3B while outperforming state-of-the-art pruning algorithms across nine zero-shot benchmarks; no multiplicative speed or memory factor is stated.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.00367",
      "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": ""
}