{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.16765", "kind": "paper", "url": "https://arxiv.org/abs/2608.16765"}, {"kind": "website", "url": "https://amuseum-whr.github.io/TraceBench"}], "authors": ["Haoran Wang", "Chaofan Ma", "Ran Yi", "Lizhuang Ma"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.16765", "source_url": "https://arxiv.org/abs/2608.16765"}, "categories": ["General AI", "Vision & 3D", "cs.CV"], "description": {"en": "Evaluates multi-reference image generation by decomposing prompts into atomic operators (Anchor, Disentangle, Apply, Compose) and scoring operator-aligned capabilities and diagnostic failure localization."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.16765", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.16765", "title": "TRACE-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.16765", "modality": null, "name": "TRACE-Bench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "Medium", "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_trace-bench_6cb6dfd0", "origin_source": "arxiv", "recognition_confidence": "0.75", "record_sha256": "5275167235d6e7d3ff3e803e038ace89435e84ff0ee93d0a9f0c8d87a574fb59", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-08-25T23:23:46.758372Z", "source_url": "https://arxiv.org/abs/2608.16765"}, "publisher": null, "released": "2026-08-17", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.16765", "source_url": "https://arxiv.org/abs/2608.16765"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-16765", "source": "claire_radar", "source_benchmark_id": "2608.16765", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Vision & 3D", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/Amuseum-WHR/TraceBench", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 14, "hfPaperUrl": "https://huggingface.co/papers/2608.16765", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-30", "state": "stale"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Formal release with project page, large case count, and acceptance at a major multimedia conference suggests moderate early attention in the multimodal generation community.", "score": 68}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-25T23:23:46.758372Z", "inputHash": "47f97fa77b1fea62329af86771fac663ce207f1fa3c9b2e0125bbbda4efa020e", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "TRACE-Bench: Decomposing and Diagnosing Multi-Reference Image Generation", "canonicalNameSource": "paper_title", "decisionReason": "TRACE-Bench has a formal name in the title, a project page with evaluation cases, and an operator-aligned scoring protocol; it provides a public path for model comparison and is accepted at ACM MM 2026.", "model": "deepseek-v4-pro", "reviewedAt": "2026-08-25T23:23:46.758372Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates multi-reference image generation by decomposing prompts into atomic operators (Anchor, Disentangle, Apply, Compose) and scoring operator-aligned capabilities and diagnostic failure localization.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "evaluation protocol evidence", "public artifact URL"], "snippet": "Building on this formulation, we construct TRACE-Bench, comprising approximately 1,600 evaluation cases across slot counts 1--8, built from 631 formula templates and around 4,000 reference images spanning diverse artistic styles and real-world subjects."}, "familyId": "bmf_4335f0096e7f", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_trace-bench_6cb6dfd0", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2608.16765", "pdf": "https://arxiv.org/pdf/2608.16765", "project": "https://amuseum-whr.github.io/TraceBench", "report": "https://arxiv.org/abs/2608.16765"}, "metrics": [], "motivation": "Despite recent advances in unified multimodal models for multi-reference image generation, existing benchmarks remain organized around predefined task types (e.g., \"subject composition\"), which are ill-suited to this combinatorial setting and lead to fragmented coverage, uncontrolled complexity, and little diagnostic value.", "name": "TRACE-Bench", "oneLine": "Evaluates multi-reference image generation by decomposing prompts into atomic operators (Anchor, Disentangle, Apply, Compose) and scoring operator-aligned capabilities and diagnostic failure localization.", "paperTitle": "TRACE-Bench: Decomposing and Diagnosing Multi-Reference Image Generation", "primaryDomain": "General AI", "publication": {"evidence": "Accepted to ACM Multimedia 2026 (ACM MM 2026)", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2608.16765", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "ACM Multimedia 2026 (ACM MM 2026)", "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "ranking": {"90d": {"confidence": "Medium", "coverage": 0.7, "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": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.47849787274907674, "value": 14}}, "confidence": "Medium", "coverage": 0.7, "rank": 542, "score": 22}, "method": "attention-ranking-v13", "rank": 542, "score": 22}}, "readiness": "Paper only", "recognitionConfidence": 0.75, "relation": "introduces", "releasedAt": "2026-08-17", "source": {"authors": ["Haoran Wang", "Chaofan Ma", "Ran Yi", "Lizhuang Ma"], "categories": ["cs.CV", "cs.AI"], "id": "2608.16765", "title": "TRACE-Bench: Decomposing and Diagnosing Multi-Reference Image Generation", "type": "arxiv", "url": "https://arxiv.org/abs/2608.16765"}, "sourceUpdatedAt": "2026-08-17T16:15:50Z", "topics": ["cs.CV"], "venueAttempts": [{"decisionRaw": "Accepted to ACM Multimedia 2026 (ACM MM 2026)", "evidence": [{"level": "author-claim", "observedAt": "2026-08-19T11:05:13.395391Z", "rawValue": "Accepted to ACM Multimedia 2026 (ACM MM 2026)", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2608.16765"}], "reviewStatus": "accepted", "venueName": "ACM Multimedia 2026 (ACM MM 2026)"}], "whyItMatters": "Provides a capability-oriented protocol that captures combinatorial complexity and enables per-operator diagnostics, addressing the fragmented coverage and limited diagnostic value of task-type-based benchmarks."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
