{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:TC4WMJGLAO4BZBANMZOBVMJQNA","short_pith_number":"pith:TC4WMJGL","schema_version":"1.0","canonical_sha256":"98b96624cb03b81c840d665c1ab1306817c5df83ec44952bbc8d931cdd8e4f7d","source":{"kind":"arxiv","id":"2605.23362","version":1},"attestation_state":"computed","paper":{"title":"Instance-Optimal Estimation with Multiple LLM Judges on a Budget","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.IT","math.IT","math.ST","stat.ML","stat.TH"],"primary_cat":"cs.LG","authors_text":"Alexandre Prouti\\`ere, Junghyun Lee, Sanghwa Kim, Se-Young Yun, Yassir Jedra","submitted_at":"2026-05-22T08:26:08Z","abstract_excerpt":"Evaluating large language models increasingly relies on LLM-as-a-judge protocols, but such evaluations remain costly: different judges have different prices and reliabilities, and the difficulty of each prompt-response pair can vary substantially. This raises a basic allocation question: under a fixed budget, how should one distribute evaluation queries across heterogeneous judges and instances to obtain the most accurate score estimates? We formalize this question as *budgeted heteroskedastic multi-judge estimation*. Given $K$ prompt-response pairs, $J$ judges with known costs, and unknown qu"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.23362","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-22T08:26:08Z","cross_cats_sorted":["cs.IT","math.IT","math.ST","stat.ML","stat.TH"],"title_canon_sha256":"157886ee4ba55b5bcbbf6334ed57396825662cd8d457360fa028982b60911ae4","abstract_canon_sha256":"5a97bf9dd44973df35faa3aa3c83cb33b9b516e8f5e0a2c0b6bc46a39683476d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-25T02:01:50.844450Z","signature_b64":"R+E9l1sp9jQ5it5DWfcssK5vvkG1tb98Jnpz/ohNBXTE6aqNV/rlYGF8RWMKyrkkVqaFTuFYmH9oznmva/geBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"98b96624cb03b81c840d665c1ab1306817c5df83ec44952bbc8d931cdd8e4f7d","last_reissued_at":"2026-05-25T02:01:50.843602Z","signature_status":"signed_v1","first_computed_at":"2026-05-25T02:01:50.843602Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Instance-Optimal Estimation with Multiple LLM Judges on a Budget","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.IT","math.IT","math.ST","stat.ML","stat.TH"],"primary_cat":"cs.LG","authors_text":"Alexandre Prouti\\`ere, Junghyun Lee, Sanghwa Kim, Se-Young Yun, Yassir Jedra","submitted_at":"2026-05-22T08:26:08Z","abstract_excerpt":"Evaluating large language models increasingly relies on LLM-as-a-judge protocols, but such evaluations remain costly: different judges have different prices and reliabilities, and the difficulty of each prompt-response pair can vary substantially. This raises a basic allocation question: under a fixed budget, how should one distribute evaluation queries across heterogeneous judges and instances to obtain the most accurate score estimates? We formalize this question as *budgeted heteroskedastic multi-judge estimation*. Given $K$ prompt-response pairs, $J$ judges with known costs, and unknown qu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.23362","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.23362/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.23362","created_at":"2026-05-25T02:01:50.843719+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.23362v1","created_at":"2026-05-25T02:01:50.843719+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.23362","created_at":"2026-05-25T02:01:50.843719+00:00"},{"alias_kind":"pith_short_12","alias_value":"TC4WMJGLAO4B","created_at":"2026-05-25T02:01:50.843719+00:00"},{"alias_kind":"pith_short_16","alias_value":"TC4WMJGLAO4BZBAN","created_at":"2026-05-25T02:01:50.843719+00:00"},{"alias_kind":"pith_short_8","alias_value":"TC4WMJGL","created_at":"2026-05-25T02:01:50.843719+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TC4WMJGLAO4BZBANMZOBVMJQNA","json":"https://pith.science/pith/TC4WMJGLAO4BZBANMZOBVMJQNA.json","graph_json":"https://pith.science/api/pith-number/TC4WMJGLAO4BZBANMZOBVMJQNA/graph.json","events_json":"https://pith.science/api/pith-number/TC4WMJGLAO4BZBANMZOBVMJQNA/events.json","paper":"https://pith.science/paper/TC4WMJGL"},"agent_actions":{"view_html":"https://pith.science/pith/TC4WMJGLAO4BZBANMZOBVMJQNA","download_json":"https://pith.science/pith/TC4WMJGLAO4BZBANMZOBVMJQNA.json","view_paper":"https://pith.science/paper/TC4WMJGL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.23362&json=true","fetch_graph":"https://pith.science/api/pith-number/TC4WMJGLAO4BZBANMZOBVMJQNA/graph.json","fetch_events":"https://pith.science/api/pith-number/TC4WMJGLAO4BZBANMZOBVMJQNA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TC4WMJGLAO4BZBANMZOBVMJQNA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TC4WMJGLAO4BZBANMZOBVMJQNA/action/storage_attestation","attest_author":"https://pith.science/pith/TC4WMJGLAO4BZBANMZOBVMJQNA/action/author_attestation","sign_citation":"https://pith.science/pith/TC4WMJGLAO4BZBANMZOBVMJQNA/action/citation_signature","submit_replication":"https://pith.science/pith/TC4WMJGLAO4BZBANMZOBVMJQNA/action/replication_record"}},"created_at":"2026-05-25T02:01:50.843719+00:00","updated_at":"2026-05-25T02:01:50.843719+00:00"}