{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.23475", "kind": "paper", "url": "https://arxiv.org/abs/2608.23475v1"}, {"kind": "website", "url": "https://anonymous.4open.science/r/StrategyBench-D53C"}], "authors": ["Jinghan Tan", "Yuanzheng Wang", "Lu Chen", "Zijun Chen", "Yuqian Wang", "Maosong Sun"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.23475", "source_url": "https://arxiv.org/abs/2608.23475"}, "categories": ["General AI", "Language & Knowledge", "cs.AI"], "description": {"en": "StrategyBench is a benchmark for evaluating explicit strategy induction in large language models. It selects strategy-inducible tasks from BIG-Bench, constructs reference strategies, and defines evaluation metrics along two dimensions: strategy quality and downstream utility."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.23475v1", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.23475v1", "title": "StrategyBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.23475", "modality": null, "name": "StrategyBench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_strategybench_2da1f1e5", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "869d0fdd8d9d88a3e6b11f970d0a23a58c7f0710d44fc119b5a3dbe101a45fe5", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-08-27T04:18:56.318436Z", "source_url": "https://arxiv.org/abs/2608.23475v1"}, "publisher": null, "released": "2026-08-24", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.23475", "source_url": "https://arxiv.org/abs/2608.23475v1"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-23475", "source": "claire_radar", "source_benchmark_id": "2608.23475", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-08-26", "attemptedAt": "2026-08-25T16:48:55.714492Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.23475", "observedAt": "2026-08-24T11:31:30.653116Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-08-25T16:48:55.714492Z", "state": "unavailable"}}, "status": "unavailable"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "The benchmark tackles a core capability in LLM adaptation and provides a rigorous evaluation, likely to gain traction in AI research communities.", "score": 50}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-27T04:22:25.189955Z", "inputHash": "76a93fcd83bd226f8d82ee00ed199e8830a3feb3ddf690dc89866d53cc102165", "model": "deepseek-v4-flash"}, "curation": {"canonicalNameEvidence": "we propose StrategyBench, which selects strategy-inducible tasks from BIG-Bench", "canonicalNameSource": "abstract", "decisionReason": "Formally named benchmark with defined metrics and a public project link.", "model": "deepseek-v4-pro", "reviewedAt": "2026-08-27T04:18:56.318436Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "StrategyBench is a benchmark for evaluating explicit strategy induction in large language models. It selects strategy-inducible tasks from BIG-Bench, constructs reference strategies, and defines evaluation metrics along two dimensions: strategy quality and downstream utility.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "To evaluate this ability, we propose StrategyBench, which selects strategy-inducible tasks from BIG-Bench, constructs reference strategies, and defines evaluation metrics along two dimensions: strategy quality and downstream utility."}, "familyId": "bmf_0e668381666c", "firstSeenAt": "2026-08-25", "heat": null, "id": "bm_strategybench_2da1f1e5", "indexedAt": "2026-08-25T16:21:05.134052Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "http://arxiv.org/abs/2608.23475v1", "pdf": "https://arxiv.org/pdf/2608.23475v1", "project": "https://anonymous.4open.science/r/StrategyBench-D53C", "report": "http://arxiv.org/abs/2608.23475v1"}, "metrics": [], "motivation": "As large language models are increasingly used in data-scarce and evolving task scenarios, few-shot in-context learning (ICL) has become a key paradigm for task adaptation.", "name": "StrategyBench", "oneLine": "StrategyBench is a benchmark for evaluating explicit strategy induction in large language models. It selects strategy-inducible tasks from BIG-Bench, constructs reference strategies, and defines evaluation metrics along two dimensions: strategy quality and downstream utility.", "paperTitle": "StrategyBench: Evaluating Explicit Strategy Induction in Large Language Models", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "http://arxiv.org/abs/2608.23475v1", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-25T16:21:05.134052Z"}, "publications": [], "publishers": [{"name": "StrategyBench Team", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://anonymous.4open.science/r/StrategyBench-D53C"}], "ranking": {"30d": {"confidence": "Low", "coverage": 0.0, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v11", "rank": null, "score": null}, "90d": {"confidence": "Low", "coverage": 0.0, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v11", "rank": null, "score": null}, "today": {"confidence": "Low", "coverage": 0.4, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}, "llmAttentionForecast": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": 2, "historyDate": "2026-08-24", "rank": null, "score": null, "windowDays": 1}, "level": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}, "llmAttentionForecast": {"percentile": 0.45, "value": 45}}, "confidence": "Low", "coverage": 0.4, "rank": 9, "score": 48}, "method": "attention-ranking-v11", "rank": 9, "score": 48}}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-24", "source": {"authors": ["Jinghan Tan", "Yuanzheng Wang", "Lu Chen", "Zijun Chen", "Yuqian Wang", "Maosong Sun"], "categories": ["cs.AI"], "id": "2608.23475", "title": "StrategyBench: Evaluating Explicit Strategy Induction in Large Language Models", "type": "arxiv", "url": "http://arxiv.org/abs/2608.23475v1"}, "sourceUpdatedAt": "2026-08-24T16:41:10Z", "topics": ["cs.AI"], "venueAttempts": [], "whyItMatters": "The benchmark addresses the gap in evaluating whether LLMs can explicitly abstract task rules from examples, which is important for adapting to data-scarce scenarios. It provides a systematic analysis of strategy induction across task variations and model configurations."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
