{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2609.04859", "kind": "paper", "url": "https://arxiv.org/abs/2609.04859"}], "authors": ["Changming Xiao", "Zhenliang Ni", "Jinhui He", "Han Shu", "Jie Hu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2609.04859", "source_url": "https://arxiv.org/abs/2609.04859"}, "categories": ["General AI", "Multimodal"], "description": {"en": "MM-IFEval-Pro evaluates vision-language models on instruction-following across Chinese and English, covering four major task categories and 24 subcategories with an average of 3.0 constraints per sample, including adversarial instruction hijacking scenarios."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2609.04859", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2609.04859", "title": "MM-IFEval-Pro"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Huawei Technologies (China)"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2609.04859", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2609.04859", "source_url": "https://openalex.org/W7211900944"}, "key": "claire-radar:2609.04859", "modality": null, "name": "MM-IFEval-Pro", "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": "false", "origin_record_id": "bm_mm-ifeval-pro_b0c09dd5", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "508dac4afa87dcc9252802884aa09f6dc154108c7fa1b4fad50e5b3967f713e9", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-name-audit-deferred", "reviewed_at": "2026-09-07T00:42:15.394147Z", "source_url": "https://arxiv.org/abs/2609.04859"}, "publisher": null, "released": "2026-09-04", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2609.04859", "source_url": "https://arxiv.org/abs/2609.04859"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2609-04859", "source": "claire_radar", "source_benchmark_id": "2609.04859", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Multimodal", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2609.04859", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "unavailable"}, "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "Instruction-following for vision-language models has broad relevance, but without released artifacts or a clear public path, early attention is expected to be moderate.", "score": 45}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-08T08:59:06.803818Z", "inputHash": "ef875d169da9adc02c426d4123762366ae0e1de29529d00fdab3c5f0b34f4a72", "model": "deepseek-v4-flash"}, "curation": {"decisionReason": "The input lacks any artifact evidence or paper PDF to confirm a formal release with an evaluation protocol.", "model": "deepseek-v4-pro", "policy": "source-evidence-v2", "reviewedAt": "2026-09-07T00:42:15.394147Z", "state": "ai-name-audit-deferred"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "MM-IFEval-Pro evaluates vision-language models on instruction-following across Chinese and English, covering four major task categories and 24 subcategories with an average of 3.0 constraints per sample, including adversarial instruction hijacking scenarios.", "displayEligible": false, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "To address these gaps, we present MM-IFEval-Pro, a multimodal instruction-following benchmark covering Chinese and English tasks as well as diverse instruction hijacking cases."}, "familyId": "bmf_a65eed5a1f3b", "firstSeenAt": "2026-09-07", "heat": null, "id": "bm_mm-ifeval-pro_b0c09dd5", "indexedAt": "2026-09-07T00:39:11.978130Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2609.04859", "pdf": "https://arxiv.org/pdf/2609.04859", "project": null, "report": "https://arxiv.org/abs/2609.04859"}, "metrics": [], "motivation": "As vision-language models (VLMs) rapidly advance in image understanding, cross-modal reasoning, and complex instruction execution, instruction-following capability has become a key indicator of their reliability and practicality.", "name": "MM-IFEval-Pro", "oneLine": "MM-IFEval-Pro evaluates vision-language models on instruction-following across Chinese and English, covering four major task categories and 24 subcategories with an average of 3.0 constraints per sample, including adversarial instruction hijacking scenarios.", "paperTitle": "MM-IFEval-Pro: A Multilingual and Attack-Resistant Benchmark for Instruction-Following in Vision-Language Models", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2609.04859", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-09-07T00:39:11.978130Z"}, "publications": [], "ranking": {"90d": {"confidence": "Low", "coverage": 0.0, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v13", "rank": null, "score": null}}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-09-04", "source": {"authors": ["Changming Xiao", "Zhenliang Ni", "Jinhui He", "Han Shu", "Jie Hu"], "categories": ["cs.AI"], "id": "2609.04859", "title": "MM-IFEval-Pro: A Multilingual and Attack-Resistant Benchmark for Instruction-Following in Vision-Language Models", "type": "arxiv", "url": "https://arxiv.org/abs/2609.04859"}, "sourceUpdatedAt": "2026-09-04T08:18:00Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "It fills gaps in multilingual coverage and adversarial safety for multimodal instruction-following benchmarks, supporting assessment of models in safety-sensitive and multilingual settings."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
