{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.29210", "kind": "paper", "url": "https://arxiv.org/abs/2608.29210"}], "authors": ["Ibrahim Mohamed Serouis", "David Jaramillo Duque"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.29210", "source_url": "https://arxiv.org/abs/2608.29210"}, "categories": ["General AI", "Vision & 3D", "cs.AI"], "description": {"en": "Evaluates text-to-image models on 1,140 prompts with 8,842 combined rules, independently varying instance count and rule constraints to measure compositional instruction following."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.29210", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.29210", "title": "Imag-Eval"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.29210", "modality": null, "name": "Imag-Eval", "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_imag-eval_e81e6086", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "3b5a6b158a1f42e30a847f5ca3525289d5423b10bf3dfcc6f3fbaf48d541e1a5", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-name-audit-deferred", "reviewed_at": "2026-09-02T05:23:11.715633Z", "source_url": "https://arxiv.org/abs/2608.29210"}, "publisher": null, "released": "2026-08-29", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.29210", "source_url": "https://arxiv.org/abs/2608.29210"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-29210", "source": "claire_radar", "source_benchmark_id": "2608.29210", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Vision & 3D", "attention": {"asOf": "2026-09-02", "attemptedAt": "2026-09-02T05:12:51.621571Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2608.29210", "observedAt": "2026-09-02T05:12:51.621571Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"observedAt": "2026-09-02T05:12:51.621571Z", "state": "fresh"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "The benchmark addresses an important T2I evaluation issue, but without immediate public artifacts, initial engagement may depend on reader follow-up to obtain materials.", "score": 58}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-02T05:23:11.715633Z", "inputHash": "5fddf717cbac1a49e87001d12f732ca045ae11ae30d88a6b56ca006b76fef7bf", "model": "deepseek-v4-pro"}, "curation": {"decisionReason": "Formally named benchmark with a clear evaluation design, but no code or data links are offered in the input beyond the paper.", "model": "deepseek-v4-pro", "policy": "source-evidence-v2", "reviewedAt": "2026-09-02T05:23:11.715633Z", "state": "ai-name-audit-deferred"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates text-to-image models on 1,140 prompts with 8,842 combined rules, independently varying instance count and rule constraints to measure compositional instruction following.", "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 limitations, we introduce Imag-Eval, a controlled benchmark designed to assess how T2I models ground compositional natural-language instructions into visual outputs."}, "familyId": "bmf_4b0929ef0aab", "firstSeenAt": "2026-09-01", "heat": null, "id": "bm_imag-eval_e81e6086", "indexedAt": "2026-09-01T06:28:29.895112Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2608.29210", "paper": "https://arxiv.org/abs/2608.29210", "pdf": "https://arxiv.org/pdf/2608.29210", "project": null, "report": "https://arxiv.org/abs/2608.29210"}, "metrics": [], "motivation": "Text-to-Image (T2I) models have recently achieved impressive visual fidelity, yet their evaluation remains constrained by benchmarks that are often difficult to interpret and insufficiently diagnostic.", "name": "Imag-Eval", "oneLine": "Evaluates text-to-image models on 1,140 prompts with 8,842 combined rules, independently varying instance count and rule constraints to measure compositional instruction following.", "paperTitle": "Imag-Eval: a language-grounded framework for interpretable Text-to-Image instruction following evaluation", "primaryDomain": "General AI", "publication": {"evidence": "Accepted to the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026)", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2608.29210", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026)", "verifiedAt": "2026-09-02T00:50:41.903960Z"}, "publications": [], "ranking": {"30d": {"confidence": "Low", "coverage": 0.3, "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": 30}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.0, "value": 0}}, "confidence": "Low", "coverage": 0.3, "rank": null, "score": 35}, "method": "attention-ranking-v13", "rank": null, "score": 35}, "90d": {"confidence": "Low", "coverage": 0.15, "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": 0.0, "value": 0}}, "confidence": "Low", "coverage": 0.15, "rank": null, "score": 43}, "method": "attention-ranking-v13", "rank": null, "score": 43}}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-29", "source": {"authors": ["Ibrahim Mohamed Serouis", "David Jaramillo Duque"], "categories": ["cs.AI"], "id": "2608.29210", "title": "Imag-Eval: a language-grounded framework for interpretable Text-to-Image instruction following evaluation", "type": "arxiv", "url": "https://arxiv.org/abs/2608.29210"}, "sourceUpdatedAt": "2026-08-29T11:50:36Z", "topics": ["cs.AI"], "venueAttempts": [{"decisionRaw": "Accepted to the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026)", "evidence": [{"level": "author-claim", "observedAt": "2026-09-02T00:50:41.903960Z", "rawValue": "Accepted to the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026)", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2608.29210"}], "reviewStatus": "accepted", "venueName": "2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026)"}], "whyItMatters": "Provides fine-grained insight into cross-modal instruction following failures by disentangling linguistic complexity and compositional difficulty."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
