{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.00909", "kind": "paper", "url": "https://arxiv.org/abs/2608.00909"}, {"id": "gh:owenfucell/finhardbench", "kind": "repo", "url": "https://github.com/owenfucell/FinHardBench"}], "authors": ["Weimin Fu", "Hejia Zhang", "Minghao Shao", "Zeng Wang", "Johann Knechtel", "Ozgur Sinanoglu", "Muhammad Shafique", "Ramesh Karri", "Xiaolong Guo"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.00909", "source_url": "https://arxiv.org/abs/2608.00909"}, "categories": ["Finance", "Language & Knowledge", "cs.CL"], "description": {"en": "A benchmark of 33 financial FPGA computing tasks that tests LLM hardware generation, system-level configuration tuning, and adaptation to specification changes, scoring both functional correctness and latency."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.00909", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.00909", "title": "FinHardBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Lehigh University", "University of California San Diego", "New York University Abu Dhabi", "New York University"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2608.00909", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2608.00909", "source_url": "https://openalex.org/W7172391354"}, "key": "claire-radar:2608.00909", "modality": null, "name": "FinHardBench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-09-25T04:40:22Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_finhardbench_a7630f10", "origin_source": "arxiv", "recognition_confidence": "0.75", "record_sha256": "e47ffe959cc24da49888480e64522b1330cb742752803fc9e1b27cc40d35bbc8", "relation": "introduces", "review_model": "claude-code-default", "review_state": "ai-reviewed", "reviewed_at": "2026-09-24T16:38:13.678431Z", "source_url": "https://arxiv.org/abs/2608.00909"}, "publisher": null, "released": "2026-08-02", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.00909", "source_url": "https://arxiv.org/abs/2608.00909"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-00909", "source": "claire_radar", "source_benchmark_id": "2608.00909", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Finance"], "area": "Language & Knowledge", "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "The paper was accepted at COLM 2026 and has a released repository, but its niche domain (financial FPGA design) limits how widely it is likely to be used.", "score": 35}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-24T16:38:13.678431Z", "inputHash": "4a9636245f20432b4e71054c62bd2991d4bb1aab9cf79d6a93e871995d9f3f08", "model": "claude-code-default"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "FinHardBench: Can LLMs Generate Latency-Aware Hardware for Financial Computing?", "canonicalNameSource": "paper_title", "decisionReason": "It is a formally named benchmark with a fixed task set and comparable correctness and timing metrics, and it is released as open source with a public code and benchmark repository.", "evaluationEvidence": {"reuse": "FinHardBench is released as an open-source benchmark.", "scoring": "models achieve 19-61% functional correctness with timing degradation up to 13.7$\\times$ on specific tasks", "task": "FinHardBench, a benchmark of 33 financial computing tasks"}, "model": "claude-code-default", "reviewedAt": "2026-09-24T16:38:13.678431Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A benchmark of 33 financial FPGA computing tasks that tests LLM hardware generation, system-level configuration tuning, and adaptation to specification changes, scoring both functional correctness and latency.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "evaluation protocol evidence", "public artifact URL"], "snippet": "FinHardBench, a benchmark of 33 financial computing tasks, is presented together with three experiments that mirror the real-world FPGA iteration cycle: generating new modules from specifications, tuning system-level configurations across a 6-stage trading pipeline, and adapting existing modules to specification changes."}, "familyId": "bmf_98692811bd52", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_finhardbench_a7630f10", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Financial Services"], "links": {"code": "https://github.com/owenfucell/FinHardBench", "data": null, "paper": "https://arxiv.org/abs/2608.00909", "pdf": "https://arxiv.org/pdf/2608.00909", "project": null, "report": "https://arxiv.org/abs/2608.00909"}, "metrics": [], "motivation": "Can large language models generate not just correct, but fast hardware?", "name": "FinHardBench", "oneLine": "A benchmark of 33 financial FPGA computing tasks that tests LLM hardware generation, system-level configuration tuning, and adaptation to specification changes, scoring both functional correctness and latency.", "paperTitle": "FinHardBench: Can LLMs Generate Latency-Aware Hardware for Financial Computing?", "primaryDomain": "Finance", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.00909", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "readiness": "Runnable", "recognitionConfidence": 0.75, "relation": "introduces", "releasedAt": "2026-08-02", "reviewLifecycle": {"state": "pending"}, "source": {"authors": ["Weimin Fu", "Hejia Zhang", "Minghao Shao", "Zeng Wang", "Johann Knechtel", "Ozgur Sinanoglu", "Muhammad Shafique", "Ramesh Karri", "Xiaolong Guo"], "categories": ["cs.CL", "cs.AR"], "id": "2608.00909", "title": "FinHardBench: Can LLMs Generate Latency-Aware Hardware for Financial Computing?", "type": "arxiv", "url": "https://arxiv.org/abs/2608.00909"}, "sourceUpdatedAt": "2026-08-02T00:30:13Z", "topics": ["cs.CL"], "venueAttempts": [], "whyItMatters": "It tests whether LLM-generated hardware is fast as well as correct, in a domain where nanoseconds of latency matter. Existing code benchmarks mostly check correctness alone."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
