{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:MEZRKB2A6YTLCP4WSBPZZJI7KI","short_pith_number":"pith:MEZRKB2A","schema_version":"1.0","canonical_sha256":"6133150740f626b13f96905f9ca51f5225647fba2098e8be8ea2d81bc7375b6b","source":{"kind":"arxiv","id":"2606.31371","version":1},"attestation_state":"computed","paper":{"title":"Calibrating the Evaluator: Does Probability Calibration Mitigate Preference Coupling in LLM Agent Feedback Loops?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Zewen Liu","submitted_at":"2026-06-30T09:03:24Z","abstract_excerpt":"When large language model (LLM) agents adapt their behavior through evaluator feedback, systematic evaluator biases propagate into the agent's learned strategy distribution - a phenomenon termed evaluator preference coupling. Prior work has documented this coupling and established a diagnostic framework (EPC) to measure it, but has not investigated whether calibration techniques can mitigate the effect. We present the first study of evaluator calibration as mitigation: applying probability calibration to the evaluator's pairwise judgments to reduce spurious preference propagation. In a control"},"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":"2606.31371","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-30T09:03:24Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"5d94d20a974a4779b73857a53856bd02f1109f8340cfb77d3f950bb0a506909d","abstract_canon_sha256":"e9f0b9d0cfad68b59b59de90043f57d7c80bb2952bf894ed02bbe9ca961e08e3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-01T01:18:00.991717Z","signature_b64":"C0z16OKvS3FTrcJgb+MbjnRWOeDjtoPYEqr5wiLHFPpimxID2GC0qydliWKQ1nkH1HxiebRX6KEhQ9wqL3j3Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6133150740f626b13f96905f9ca51f5225647fba2098e8be8ea2d81bc7375b6b","last_reissued_at":"2026-07-01T01:18:00.991263Z","signature_status":"signed_v1","first_computed_at":"2026-07-01T01:18:00.991263Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Calibrating the Evaluator: Does Probability Calibration Mitigate Preference Coupling in LLM Agent Feedback Loops?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Zewen Liu","submitted_at":"2026-06-30T09:03:24Z","abstract_excerpt":"When large language model (LLM) agents adapt their behavior through evaluator feedback, systematic evaluator biases propagate into the agent's learned strategy distribution - a phenomenon termed evaluator preference coupling. Prior work has documented this coupling and established a diagnostic framework (EPC) to measure it, but has not investigated whether calibration techniques can mitigate the effect. We present the first study of evaluator calibration as mitigation: applying probability calibration to the evaluator's pairwise judgments to reduce spurious preference propagation. In a control"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.31371","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/2606.31371/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":"2606.31371","created_at":"2026-07-01T01:18:00.991327+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.31371v1","created_at":"2026-07-01T01:18:00.991327+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.31371","created_at":"2026-07-01T01:18:00.991327+00:00"},{"alias_kind":"pith_short_12","alias_value":"MEZRKB2A6YTL","created_at":"2026-07-01T01:18:00.991327+00:00"},{"alias_kind":"pith_short_16","alias_value":"MEZRKB2A6YTLCP4W","created_at":"2026-07-01T01:18:00.991327+00:00"},{"alias_kind":"pith_short_8","alias_value":"MEZRKB2A","created_at":"2026-07-01T01:18:00.991327+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/MEZRKB2A6YTLCP4WSBPZZJI7KI","json":"https://pith.science/pith/MEZRKB2A6YTLCP4WSBPZZJI7KI.json","graph_json":"https://pith.science/api/pith-number/MEZRKB2A6YTLCP4WSBPZZJI7KI/graph.json","events_json":"https://pith.science/api/pith-number/MEZRKB2A6YTLCP4WSBPZZJI7KI/events.json","paper":"https://pith.science/paper/MEZRKB2A"},"agent_actions":{"view_html":"https://pith.science/pith/MEZRKB2A6YTLCP4WSBPZZJI7KI","download_json":"https://pith.science/pith/MEZRKB2A6YTLCP4WSBPZZJI7KI.json","view_paper":"https://pith.science/paper/MEZRKB2A","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.31371&json=true","fetch_graph":"https://pith.science/api/pith-number/MEZRKB2A6YTLCP4WSBPZZJI7KI/graph.json","fetch_events":"https://pith.science/api/pith-number/MEZRKB2A6YTLCP4WSBPZZJI7KI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MEZRKB2A6YTLCP4WSBPZZJI7KI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MEZRKB2A6YTLCP4WSBPZZJI7KI/action/storage_attestation","attest_author":"https://pith.science/pith/MEZRKB2A6YTLCP4WSBPZZJI7KI/action/author_attestation","sign_citation":"https://pith.science/pith/MEZRKB2A6YTLCP4WSBPZZJI7KI/action/citation_signature","submit_replication":"https://pith.science/pith/MEZRKB2A6YTLCP4WSBPZZJI7KI/action/replication_record"}},"created_at":"2026-07-01T01:18:00.991327+00:00","updated_at":"2026-07-01T01:18:00.991327+00:00"}