{ "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": "2610.02882", "title": "DyRA: Dynamic Residual Approximation for Efficient Matrix Multiplication in DNNs", "summary": "DyRA replaces dense weight matrices with structured low-rank approximations augmented by an input-dependent, dynamically computed low-rank correction of the output residual error. The cost is the extra runtime computation of the residual factors per input, added implementation complexity in the matmul path, and a per-model fitting or calibration step before deployment. It can fail when the correction overhead erodes the speedup on layers or batch sizes where it dominates, when the dynamic factors are estimated under distribution shift relative to the calibration data, or when target models lack accessible weight matrices amenable to structured factorization.", "rationale": "The method offers a validated accuracy-efficiency improvement applicable to pretrained models on a single GPU, but the speedup and correction quality are workload-dependent and require per-model calibration, so small-scale validation on the reader's actual models is warranted before production adoption.", "technique": "Input-adaptive residual correction for low-rank matrix multiplication", "family": "arquitetura_eficiente", "recommendation": "testar", "published": "2026-10-02", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.02882", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "DyRA replaces dense weight matrices with structured low-rank approximations augmented by an input-dependent, dynamically computed low-rank correction of the output residual error. The cost is the extra runtime computation of the residual factors per input, added implementation complexity in the matmul path, and a per-model fitting or calibration step before deployment. It can fail when the correction overhead erodes the speedup on layers or batch sizes where it dominates, when the dynamic factors are estimated under distribution shift relative to the calibration data, or when target models lack accessible weight matrices amenable to structured factorization.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.02882", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "DyRA achieves a 1.5x end-to-end GPU speedup for DINOv3 while reducing accuracy degradation by more than 3x relative to weight-only structured approximation baselines.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.02882", "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": "" }