{ "schema_version": 1, "builder_plan": { "scenario": "A product assistant retrieves plausible documents but gives unsupported answers.", "change": "Change one retrieval or ranking step while keeping the document snapshot fixed.", "baseline": "Your current retrieval pipeline with identical questions and answer model.", "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": "Supported answers, retrieval misses, correct abstentions, index cost, and response latency.", "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": "2610.02875", "title": "Query-aware routing for Cross-lingual performance gains in Encoders", "summary": "Replaces retraining a multilingual encoder or re-embedding the document index with a small query-only LoRA adapter trained against frozen embeddings, plus deterministic routing that sends cross-language queries to the adapter and same-language queries to the original encoder. Costs one small adapter to train and host (LoRA on a 1B encoder fits easily on a 24 GB GPU), a routing check per query, and no document re-indexing; same-language quality is preserved by construction. Can fail if routing misclassifies query language, if gains do not transfer beyond the evaluated language pairs or financial domain, or if adapter and index drift out of sync when the base encoder is later updated.", "rationale": "A constrained-infrastructure team doing multilingual retrieval can adopt this directly: it requires only a small adapter trained on one GPU, preserves the existing index and same-language behavior, and shows a measured cross-lingual nDCG gain in every evaluated direction.", "technique": "Query-only LoRA adapter with language-based routing", "family": "recuperacao_e_rag", "recommendation": "adotar", "published": "2026-10-02", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.02875", "independent_implementations": 0, "total_implementations": 1, "stars": 1, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "Replaces retraining a multilingual encoder or re-embedding the document index with a small query-only LoRA adapter trained against frozen embeddings, plus deterministic routing that sends cross-language queries to the adapter and same-language queries to the original encoder. Costs one small adapter to train and host (LoRA on a 1B encoder fits easily on a 24 GB GPU), a routing check per query, and no document re-indexing; same-language quality is preserved by construction. Can fail if routing misclassifies query language, if gains do not transfer beyond the evaluated language pairs or financial domain, or if adapter and index drift out of sync when the base encoder is later updated.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.02875", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "Average nDCG@10 improves from 0.241 to 0.291 (20.9% relative) across six English/Finnish/Swedish cross-lingual retrieval directions on a sampled financial benchmark, with same-language performance preserved by routing.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.02875", "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": "" }