{ "schema_version": 1, "builder_plan": { "scenario": "You already need task-specific training and want to reduce its memory or compute cost.", "change": "Change only the training technique on a small controlled run after checking prerequisites.", "baseline": "Your current training recipe with the same model, data split, and evaluation.", "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": "Held-out quality, peak memory, training time, and full compute cost to the target quality.", "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": "2610.19700", "title": "ReGRA: Residual-Gradient Adaptation for Complementary Low-Rank Fine-Tuning", "summary": "ReGRA replaces a single-stage LoRA update with two complementary low-rank stages: a standard first-stage adapter, then a second adapter initialized from residual task gradients projected away from the first-stage subspace, with an orthogonality regularizer. It costs an extra fine-tuning stage, a calibration set, and gradient-estimation compute, but merges into the base weights with no inference overhead. It can fail if the calibration set is unrepresentative, if the residual-gradient estimate is noisy on small datasets, or if the two-stage pipeline's added hyperparameters are not retuned for the target task.", "rationale": "The method fits a 24 GB GPU budget and may extract more quality from a fixed rank, but the reported gains are rank-relative and the two-stage procedure adds complexity, so workload-specific validation is warranted before production adoption.", "technique": "Two-stage LoRA with residual-gradient initialization", "family": "treino_eficiente", "recommendation": "testar", "published": "2026-10-14", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.19700", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": 0, "claims": [ { "claim_id": "claim-01", "statement": "ReGRA replaces a single-stage LoRA update with two complementary low-rank stages: a standard first-stage adapter, then a second adapter initialized from residual task gradients projected away from the first-stage subspace, with an orthogonality regularizer. It costs an extra fine-tuning stage, a calibration set, and gradient-estimation compute, but merges into the base weights with no inference overhead. It can fail if the calibration set is unrepresentative, if the residual-gradient estimate is noisy on small datasets, or if the two-stage pipeline's added hyperparameters are not retuned for the target task.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.19700", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "ReGRA outperforms alternatives at the same total rank and matches LoRA adapters with twice the rank across diverse benchmarks; a rank-16 diagnostic shows top singular components carry most of the useful adaptation.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.19700", "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": "" }