{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:HB7NV4PM3Q3DGI2QIEPK32MXNK","short_pith_number":"pith:HB7NV4PM","schema_version":"1.0","canonical_sha256":"387edaf1ecdc36332350411eade9976abf3234fa7b908b2c41f8688f9b5dbe34","source":{"kind":"arxiv","id":"2510.05342","version":2},"attestation_state":"computed","paper":{"title":"Margin Adaptive DPO: Leveraging Reward Model for Granular Control in Preference Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hyung Gyu Rho","submitted_at":"2025-10-06T20:09:37Z","abstract_excerpt":"Direct Preference Optimization (DPO) has emerged as a simple and effective method for aligning large language models. However, its reliance on a fixed temperature parameter leads to suboptimal training on diverse preference data, causing overfitting on easy examples and under-learning from informative ones. Recent methods have emerged to counter this. While IPO addresses general overfitting, its uniform regularization can be overly conservative. The more targeted approach of $\\beta$-DPO suffers from its own limitations: its batch-level adaptation applies a single, compromised temperature to mi"},"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":"2510.05342","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-10-06T20:09:37Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2e70e7b8b5d3a1eaadc27c6b3b54e4b75436b168500ef995c67fe57900034c60","abstract_canon_sha256":"92774fafed989d57ad664c5ddc55f6f3dea3f6abe223de8c085c32410d5a0541"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:04:48.192422Z","signature_b64":"odH7B2kdv0F3oJ3t/iPP/g4oEYOCBwWTtxcrseO8GNO8dsU6ghqDL3Y7zelDF3uVnwbWKqS8oz5f1UcW3HUsCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"387edaf1ecdc36332350411eade9976abf3234fa7b908b2c41f8688f9b5dbe34","last_reissued_at":"2026-06-02T02:04:48.191971Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:04:48.191971Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Margin Adaptive DPO: Leveraging Reward Model for Granular Control in Preference Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hyung Gyu Rho","submitted_at":"2025-10-06T20:09:37Z","abstract_excerpt":"Direct Preference Optimization (DPO) has emerged as a simple and effective method for aligning large language models. However, its reliance on a fixed temperature parameter leads to suboptimal training on diverse preference data, causing overfitting on easy examples and under-learning from informative ones. Recent methods have emerged to counter this. While IPO addresses general overfitting, its uniform regularization can be overly conservative. The more targeted approach of $\\beta$-DPO suffers from its own limitations: its batch-level adaptation applies a single, compromised temperature to mi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.05342","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/2510.05342/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":"2510.05342","created_at":"2026-06-02T02:04:48.192024+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.05342v2","created_at":"2026-06-02T02:04:48.192024+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.05342","created_at":"2026-06-02T02:04:48.192024+00:00"},{"alias_kind":"pith_short_12","alias_value":"HB7NV4PM3Q3D","created_at":"2026-06-02T02:04:48.192024+00:00"},{"alias_kind":"pith_short_16","alias_value":"HB7NV4PM3Q3DGI2Q","created_at":"2026-06-02T02:04:48.192024+00:00"},{"alias_kind":"pith_short_8","alias_value":"HB7NV4PM","created_at":"2026-06-02T02:04:48.192024+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2604.17299","citing_title":"Cat-DPO: Category-Adaptive Safety Alignment","ref_index":35,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HB7NV4PM3Q3DGI2QIEPK32MXNK","json":"https://pith.science/pith/HB7NV4PM3Q3DGI2QIEPK32MXNK.json","graph_json":"https://pith.science/api/pith-number/HB7NV4PM3Q3DGI2QIEPK32MXNK/graph.json","events_json":"https://pith.science/api/pith-number/HB7NV4PM3Q3DGI2QIEPK32MXNK/events.json","paper":"https://pith.science/paper/HB7NV4PM"},"agent_actions":{"view_html":"https://pith.science/pith/HB7NV4PM3Q3DGI2QIEPK32MXNK","download_json":"https://pith.science/pith/HB7NV4PM3Q3DGI2QIEPK32MXNK.json","view_paper":"https://pith.science/paper/HB7NV4PM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.05342&json=true","fetch_graph":"https://pith.science/api/pith-number/HB7NV4PM3Q3DGI2QIEPK32MXNK/graph.json","fetch_events":"https://pith.science/api/pith-number/HB7NV4PM3Q3DGI2QIEPK32MXNK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HB7NV4PM3Q3DGI2QIEPK32MXNK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HB7NV4PM3Q3DGI2QIEPK32MXNK/action/storage_attestation","attest_author":"https://pith.science/pith/HB7NV4PM3Q3DGI2QIEPK32MXNK/action/author_attestation","sign_citation":"https://pith.science/pith/HB7NV4PM3Q3DGI2QIEPK32MXNK/action/citation_signature","submit_replication":"https://pith.science/pith/HB7NV4PM3Q3DGI2QIEPK32MXNK/action/replication_record"}},"created_at":"2026-06-02T02:04:48.192024+00:00","updated_at":"2026-06-02T02:04:48.192024+00:00"}