{ "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": "2610.30041", "title": "Robust Soft Skill Adaptation via Behavioral Localization in Tool-Using Agents", "summary": "The method replaces supervising soft prompts on entire successful trajectories (or passing long text skills at inference) with supervision restricted to trajectory positions where the text skill both raises the likelihood of successful tokens and shifts the next-token distribution. It costs extra frozen-model forward passes, with and without the skill, to score every position before training, and adds a localization pipeline on top of standard soft-prompt training. It can fail when skill effects are diffuse rather than concentrated, when the localization threshold or top-k fraction is miscalibrated for a new domain, or when gains measured on a single benchmark (SpreadsheetBench, one frozen model) do not transfer to the reader's tasks.", "rationale": "The approach is feasible on constrained infrastructure since it requires only inference on a frozen model plus training a short soft prompt, but the reported gains come from a single benchmark and model, so workload-specific validation is warranted before production adoption.", "technique": "Behavioral localization for soft-prompt supervision", "family": "treino_eficiente", "recommendation": "testar", "published": "2026-10-21", "as_of": "2026-10-22", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.30041", "independent_implementations": 1, "total_implementations": 2, "stars": 132, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "The method replaces supervising soft prompts on entire successful trajectories (or passing long text skills at inference) with supervision restricted to trajectory positions where the text skill both raises the likelihood of successful tokens and shifts the next-token distribution. It costs extra frozen-model forward passes, with and without the skill, to score every position before training, and adds a localization pipeline on top of standard soft-prompt training. It can fail when skill effects are diffuse rather than concentrated, when the localization threshold or top-k fraction is miscalibrated for a new domain, or when gains measured on a single benchmark (SpreadsheetBench, one frozen model) do not transfer to the reader's tasks.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.30041", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "Task-specific soft prompts trained on the top 10% localized positions reach 57.38% success on SpreadsheetBench, outperforming both the full text skill and the on-policy OPCD baseline; in the shared-prompt setting selective supervision improves success from 30.36% (full trajectory) to 36.79%.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.30041", "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": "" }