{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6ITC2NIOLKRDKOWJYGETFJTGVV","short_pith_number":"pith:6ITC2NIO","schema_version":"1.0","canonical_sha256":"f2262d350e5aa2353ac9c18932a666ad7bc1f578492671ccf3c6ac5f9606abf3","source":{"kind":"arxiv","id":"2512.03019","version":2},"attestation_state":"computed","paper":{"title":"Distribution-Calibrated Inference Time Compute for Thinking LLM-as-a-Judge","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Firas Trabelsi, Hamid Dadkhahi, Juraj Juraska, Mehdi Mirzazadeh, Parker Riley","submitted_at":"2025-12-02T18:46:47Z","abstract_excerpt":"Thinking Large Language Models (LLMs) used as judges for pairwise preferences remain noisy at the single-sample level, and common aggregation rules (majority vote, soft self-consistency, or instruction-based self-aggregation) are inconsistent when ties are allowed. We study inference-time compute (ITC) for evaluators that generate n independent thinking--rating samples per item, and propose a principled, distribution-calibrated aggregation scheme. Our method models three-way preferences with a Bradley-Terry-Davidson formulation on rating counts, leveraging both polarity (margin among non-ties)"},"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":"2512.03019","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-12-02T18:46:47Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"24b792f93236eb3073712ede427c71ab8bafd9883504680df0632e9d528131d3","abstract_canon_sha256":"ff17e301b4ffc9f39db33e4acad6f08ec12135dd490ed93e859d13689f37e474"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-03T01:05:07.912551Z","signature_b64":"vSvPssUBelnPJ5LoRmppxnjZgHGqCVgZQQI30uo6+YojDLM9erL3pTmEGGDw9HnBwvNyNDoJbvdcK0MmLBPVDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f2262d350e5aa2353ac9c18932a666ad7bc1f578492671ccf3c6ac5f9606abf3","last_reissued_at":"2026-06-03T01:05:07.912028Z","signature_status":"signed_v1","first_computed_at":"2026-06-03T01:05:07.912028Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Distribution-Calibrated Inference Time Compute for Thinking LLM-as-a-Judge","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Firas Trabelsi, Hamid Dadkhahi, Juraj Juraska, Mehdi Mirzazadeh, Parker Riley","submitted_at":"2025-12-02T18:46:47Z","abstract_excerpt":"Thinking Large Language Models (LLMs) used as judges for pairwise preferences remain noisy at the single-sample level, and common aggregation rules (majority vote, soft self-consistency, or instruction-based self-aggregation) are inconsistent when ties are allowed. We study inference-time compute (ITC) for evaluators that generate n independent thinking--rating samples per item, and propose a principled, distribution-calibrated aggregation scheme. Our method models three-way preferences with a Bradley-Terry-Davidson formulation on rating counts, leveraging both polarity (margin among non-ties)"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.03019","kind":"arxiv","version":2},"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/2512.03019/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":"2512.03019","created_at":"2026-06-03T01:05:07.912095+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.03019v2","created_at":"2026-06-03T01:05:07.912095+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.03019","created_at":"2026-06-03T01:05:07.912095+00:00"},{"alias_kind":"pith_short_12","alias_value":"6ITC2NIOLKRD","created_at":"2026-06-03T01:05:07.912095+00:00"},{"alias_kind":"pith_short_16","alias_value":"6ITC2NIOLKRDKOWJ","created_at":"2026-06-03T01:05:07.912095+00:00"},{"alias_kind":"pith_short_8","alias_value":"6ITC2NIO","created_at":"2026-06-03T01:05:07.912095+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.09702","citing_title":"Calibrate, Don't Curate: Label-Efficient Estimation from Noisy LLM Judges","ref_index":3,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6ITC2NIOLKRDKOWJYGETFJTGVV","json":"https://pith.science/pith/6ITC2NIOLKRDKOWJYGETFJTGVV.json","graph_json":"https://pith.science/api/pith-number/6ITC2NIOLKRDKOWJYGETFJTGVV/graph.json","events_json":"https://pith.science/api/pith-number/6ITC2NIOLKRDKOWJYGETFJTGVV/events.json","paper":"https://pith.science/paper/6ITC2NIO"},"agent_actions":{"view_html":"https://pith.science/pith/6ITC2NIOLKRDKOWJYGETFJTGVV","download_json":"https://pith.science/pith/6ITC2NIOLKRDKOWJYGETFJTGVV.json","view_paper":"https://pith.science/paper/6ITC2NIO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.03019&json=true","fetch_graph":"https://pith.science/api/pith-number/6ITC2NIOLKRDKOWJYGETFJTGVV/graph.json","fetch_events":"https://pith.science/api/pith-number/6ITC2NIOLKRDKOWJYGETFJTGVV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6ITC2NIOLKRDKOWJYGETFJTGVV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6ITC2NIOLKRDKOWJYGETFJTGVV/action/storage_attestation","attest_author":"https://pith.science/pith/6ITC2NIOLKRDKOWJYGETFJTGVV/action/author_attestation","sign_citation":"https://pith.science/pith/6ITC2NIOLKRDKOWJYGETFJTGVV/action/citation_signature","submit_replication":"https://pith.science/pith/6ITC2NIOLKRDKOWJYGETFJTGVV/action/replication_record"}},"created_at":"2026-06-03T01:05:07.912095+00:00","updated_at":"2026-06-03T01:05:07.912095+00:00"}