{ "schema_version": 1, "builder_plan": { "scenario": "Start with one repeated product task that has a measurable failure or cost.", "change": "Read the methods and identify one change you can isolate in that workflow.", "baseline": "Your current implementation on the same held-out tasks.", "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": "Successful outcomes, response time, human rework, and total cost per successful task.", "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": "2610.20253", "title": "Repairing Speech Access After Text Updates with Frozen Affine Fields", "summary": "AFFIRM replaces retraining or refitting the speech-language model after a text update with a frozen, detachable affine correction applied to speech hidden states in a single pass. It costs an operator-training phase over paired speech and text states plus a small per-query conditioning computation, keeping the base checkpoint untouched. It can fail to outperform simpler baselines: on balanced synthetic PopQA it showed no established advantage over SALSA steering or low-rank ReFT, and steering methods can degrade when transferred across domains, as the matched SALSA variant did.", "rationale": "The method is cheap to run under constrained infrastructure and shows positive zero-shot transfer to recorded speech, but its advantage over low-rank ReFT is unproven, so a team operating speech-capable LLMs should validate it against ReFT on their own workload before adopting.", "technique": "Frozen affine steering fields for speech access repair", "family": "outro", "recommendation": "testar", "published": "2026-10-14", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.20253", "independent_implementations": 0, "total_implementations": 1, "stars": 0, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "AFFIRM replaces retraining or refitting the speech-language model after a text update with a frozen, detachable affine correction applied to speech hidden states in a single pass. It costs an operator-training phase over paired speech and text states plus a small per-query conditioning computation, keeping the base checkpoint untouched. It can fail to outperform simpler baselines: on balanced synthetic PopQA it showed no established advantage over SALSA steering or low-rank ReFT, and steering methods can degrade when transferred across domains, as the matched SALSA variant did.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.20253", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "On recorded SLUE-SQA-5 questions, a PopQA-fitted AFFIRM operator transferred without refitting and improved ROUGE-L by 0.0941 over its paired no-repair baseline, while the matched SALSA variant fell 0.0707 below baseline; on synthetic PopQA no clear advantage over SALSA or ReFT was established.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.20253", "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": "" }