{"record": {"aliases": ["KVDiagnosis: A Diagnostic Benchmark for KV-Cache Compression in Long-Context Language Models"], "artifacts": [{"id": "arxiv:2608.09412", "kind": "paper", "url": "https://arxiv.org/abs/2608.09412"}, {"id": "gh:chosenqc/kvdiagnosis", "kind": "repo", "url": "https://github.com/ChosenQC/KVDiagnosis"}], "authors": ["Chen Qiu", "Ziwu Liu", "Chao Fei", "Guozhong Li", "Panos Kalnis"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.09412", "source_url": "https://arxiv.org/abs/2608.09412"}, "categories": ["General AI", "Language & Knowledge", "Long Context"], "description": {"en": "Diagnoses KV-cache compression failures in long-context language models using paired runs and cache, likelihood, attention, and decoding measurements."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.09412", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.09412", "title": "KVDiagnosis"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["King Abdullah University of Science and Technology"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2608.09412", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2608.09412", "source_url": "https://openalex.org/W7202179078"}, "key": "claire-radar:2608.09412", "modality": null, "name": "KVDiagnosis", "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_kvdiagnosis-a-diagnostic-benchmark-for-kv-_86d87177", "origin_source": "arxiv", "recognition_confidence": "0.7", "record_sha256": "02ceeebef14a2e1b9850b5cb418cb6b8324a6fd1a437e0e7aad6d0034fedb78b", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-08-27T04:14:52.836173Z", "source_url": "https://arxiv.org/abs/2608.09412"}, "publisher": null, "released": "2026-08-10", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.09412", "source_url": "https://arxiv.org/abs/2608.09412"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-09412", "source": "claire_radar", "source_benchmark_id": "2608.09412", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["KVDiagnosis: A Diagnostic Benchmark for KV-Cache Compression in Long-Context Language Models"], "annotation": "Mixed", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/ChosenQC/KVDiagnosis", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.09412", "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", "state": "unavailable"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Targets a timely problem in long-context LLM efficiency and provides a comprehensive public repository with detailed diagnostics.", "score": 80}, "capabilities": [], "confidence": "Medium", "construction": "Aggregate Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-27T04:14:52.836173Z", "inputHash": "558a3f8d973e6b212e25ac4e45fdd7f843ecb729dd8197c129b32a4952dadbe5", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "We present KVDiagnosis, a diagnostic dataset and benchmark", "canonicalNameSource": "abstract", "decisionReason": "Released with code and data, and defines a diagnostic protocol with fixed rules and metrics; intended for ongoing model comparison.", "model": "deepseek-v4-pro", "reviewedAt": "2026-08-27T04:14:52.836173Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Diagnoses KV-cache compression failures in long-context language models using paired runs and cache, likelihood, attention, and decoding measurements.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["benchmark term in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We present KVDiagnosis, a diagnostic dataset and benchmark with three contributions."}, "familyId": "bmf_00adefbae7c6", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_kvdiagnosis-a-diagnostic-benchmark-for-kv-_86d87177", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/ChosenQC/KVDiagnosis", "data": null, "paper": "https://arxiv.org/abs/2608.09412", "pdf": "https://arxiv.org/pdf/2608.09412", "project": null, "report": "https://arxiv.org/abs/2608.09412"}, "metrics": [], "motivation": "KV-cache compression reduces long-context memory, but aggregate task scores reveal neither which correct executions fail nor why.", "name": "KVDiagnosis", "oneLine": "Diagnoses KV-cache compression failures in long-context language models using paired runs and cache, likelihood, attention, and decoding measurements.", "paperTitle": "KVDiagnosis: A Diagnostic Benchmark for KV-Cache Compression in Long-Context Language Models", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.09412", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "publishers": [{"name": "ChosenQC", "organizationType": "community", "role": "benchmark-publisher", "sourceUrl": "https://github.com/ChosenQC/KVDiagnosis"}], "ranking": {"90d": {"confidence": "Low", "coverage": 0.55, "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": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 528, "score": 23}, "method": "attention-ranking-v13", "rank": 528, "score": 23}}, "readiness": "Runnable", "recognitionConfidence": 0.7, "relation": "introduces", "releasedAt": "2026-08-10", "source": {"authors": ["Chen Qiu", "Ziwu Liu", "Chao Fei", "Guozhong Li", "Panos Kalnis"], "categories": ["cs.AI"], "id": "2608.09412", "title": "KVDiagnosis: A Diagnostic Benchmark for KV-Cache Compression in Long-Context Language Models", "type": "arxiv", "url": "https://arxiv.org/abs/2608.09412"}, "sourceUpdatedAt": "2026-08-10T10:38:05Z", "topics": ["Long Context"], "venueAttempts": [], "whyItMatters": "Provides failure-focused diagnostics that reveal specific causes of compression errors, moving beyond aggregate task scores."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
