{ "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": "2610.10349", "title": "AutoAdapt: Automatic Domain Discovery Enables Low-Cost Extensibility", "summary": "It replaces retraining a monolithic LoRA (or full model) on all data whenever a new domain arrives, substituting independently trained per-domain LoRA adapters with parameter-free routing over automatically discovered domains. The cost is training and storing one small adapter per discovered domain, plus reliance on an unsupervised domain-discovery step whose quality determines adapter usefulness; per the paper there is no aggregate accuracy loss, but parity rather than improvement is the ceiling. It can fail if latent domain discovery missegments data (too coarse, too fine, or drifting over time), if routing selects the wrong adapter for mixed-domain inputs, or if a genuinely new domain is not recognized and routed correctly.", "rationale": "Modular per-domain LoRA adapters are feasible on a single 24 GB GPU and eliminate retraining when domains are added, but the value of automatic discovery and parameter-free routing versus simpler manual domain labeling must be validated on the reader's actual workload.", "technique": "Automatic domain discovery with per-domain LoRA adapters", "family": "treino_eficiente", "recommendation": "testar", "published": "2026-10-07", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.10349", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "It replaces retraining a monolithic LoRA (or full model) on all data whenever a new domain arrives, substituting independently trained per-domain LoRA adapters with parameter-free routing over automatically discovered domains. The cost is training and storing one small adapter per discovered domain, plus reliance on an unsupervised domain-discovery step whose quality determines adapter usefulness; per the paper there is no aggregate accuracy loss, but parity rather than improvement is the ceiling. It can fail if latent domain discovery missegments data (too coarse, too fine, or drifting over time), if routing selects the wrong adapter for mixed-domain inputs, or if a genuinely new domain is not recognized and routed correctly.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.10349", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "Achieves parity with a LoRA adapter trained on all domains across 14 benchmarks and GPT-4o pairwise judgements, without full-model retraining.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.10349", "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": "" }