{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:CMXX5V7OEHTLOFPE5Z54JN5TSL","short_pith_number":"pith:CMXX5V7O","schema_version":"1.0","canonical_sha256":"132f7ed7ee21e6b715e4ee7bc4b7b392fb81b8e4246142d798392668faa38136","source":{"kind":"arxiv","id":"2602.04909","version":3},"attestation_state":"computed","paper":{"title":"Learning Where It Matters: Geometric Anchoring for Robust Preference Alignment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Ji-Hoon Kim, Jongsuk Kim, Youngjae Cho","submitted_at":"2026-02-04T00:40:21Z","abstract_excerpt":"Direct Preference Optimization (DPO) and related methods align large language models from pairwise preferences by regularizing updates against a fixed reference policy. As the policy drifts, a static reference, however, can become increasingly miscalibrated, leading to distributional mismatch and amplifying spurious preference signals under noisy supervision. Conversely, reference-free variants avoid mismatch but often suffer from unconstrained reward drift. We propose Geometric Anchor Preference Optimization (GAPO), which replaces the fixed reference with a dynamic, geometry-aware anchor: an "},"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":"2602.04909","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-04T00:40:21Z","cross_cats_sorted":[],"title_canon_sha256":"1d3d8c46518355407869ab680b699240b9c0e9ae4dc866d41110643b2e5ed435","abstract_canon_sha256":"eee7e5151eee0f7e9467f8b825a826140da8b6e8fdc85b55ddcffe54293d1116"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:33.564762Z","signature_b64":"PYkrYZUPGHlBMJ3sJijAPbaX8cGio+dsWlB6fmHZIsKjKbhdkI9WiGP1AK1LH87YiNma9j5/Od8mWo7AeqNLCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"132f7ed7ee21e6b715e4ee7bc4b7b392fb81b8e4246142d798392668faa38136","last_reissued_at":"2026-05-20T00:00:33.564166Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:33.564166Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Where It Matters: Geometric Anchoring for Robust Preference Alignment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Ji-Hoon Kim, Jongsuk Kim, Youngjae Cho","submitted_at":"2026-02-04T00:40:21Z","abstract_excerpt":"Direct Preference Optimization (DPO) and related methods align large language models from pairwise preferences by regularizing updates against a fixed reference policy. As the policy drifts, a static reference, however, can become increasingly miscalibrated, leading to distributional mismatch and amplifying spurious preference signals under noisy supervision. Conversely, reference-free variants avoid mismatch but often suffer from unconstrained reward drift. We propose Geometric Anchor Preference Optimization (GAPO), which replaces the fixed reference with a dynamic, geometry-aware anchor: an "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.04909","kind":"arxiv","version":3},"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/2602.04909/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":"2602.04909","created_at":"2026-05-20T00:00:33.564257+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.04909v3","created_at":"2026-05-20T00:00:33.564257+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.04909","created_at":"2026-05-20T00:00:33.564257+00:00"},{"alias_kind":"pith_short_12","alias_value":"CMXX5V7OEHTL","created_at":"2026-05-20T00:00:33.564257+00:00"},{"alias_kind":"pith_short_16","alias_value":"CMXX5V7OEHTLOFPE","created_at":"2026-05-20T00:00:33.564257+00:00"},{"alias_kind":"pith_short_8","alias_value":"CMXX5V7O","created_at":"2026-05-20T00:00:33.564257+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/CMXX5V7OEHTLOFPE5Z54JN5TSL","json":"https://pith.science/pith/CMXX5V7OEHTLOFPE5Z54JN5TSL.json","graph_json":"https://pith.science/api/pith-number/CMXX5V7OEHTLOFPE5Z54JN5TSL/graph.json","events_json":"https://pith.science/api/pith-number/CMXX5V7OEHTLOFPE5Z54JN5TSL/events.json","paper":"https://pith.science/paper/CMXX5V7O"},"agent_actions":{"view_html":"https://pith.science/pith/CMXX5V7OEHTLOFPE5Z54JN5TSL","download_json":"https://pith.science/pith/CMXX5V7OEHTLOFPE5Z54JN5TSL.json","view_paper":"https://pith.science/paper/CMXX5V7O","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.04909&json=true","fetch_graph":"https://pith.science/api/pith-number/CMXX5V7OEHTLOFPE5Z54JN5TSL/graph.json","fetch_events":"https://pith.science/api/pith-number/CMXX5V7OEHTLOFPE5Z54JN5TSL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CMXX5V7OEHTLOFPE5Z54JN5TSL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CMXX5V7OEHTLOFPE5Z54JN5TSL/action/storage_attestation","attest_author":"https://pith.science/pith/CMXX5V7OEHTLOFPE5Z54JN5TSL/action/author_attestation","sign_citation":"https://pith.science/pith/CMXX5V7OEHTLOFPE5Z54JN5TSL/action/citation_signature","submit_replication":"https://pith.science/pith/CMXX5V7OEHTLOFPE5Z54JN5TSL/action/replication_record"}},"created_at":"2026-05-20T00:00:33.564257+00:00","updated_at":"2026-05-20T00:00:33.564257+00:00"}