{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:EIUNGFN63U756SOQVID6C7AI6X","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"7ea2787c4e7da787feb91acea81b32f5c5264898cdefbff30c0ec13b30474a16","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-03-23T16:20:59Z","title_canon_sha256":"a63c3e94f31705cb0ab702c2bd7b63449edcd63dc40f5f703f618cda8f372b34"},"schema_version":"1.0","source":{"id":"2503.18130","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.18130","created_at":"2026-07-05T10:38:03Z"},{"alias_kind":"arxiv_version","alias_value":"2503.18130v1","created_at":"2026-07-05T10:38:03Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.18130","created_at":"2026-07-05T10:38:03Z"},{"alias_kind":"pith_short_12","alias_value":"EIUNGFN63U75","created_at":"2026-07-05T10:38:03Z"},{"alias_kind":"pith_short_16","alias_value":"EIUNGFN63U756SOQ","created_at":"2026-07-05T10:38:03Z"},{"alias_kind":"pith_short_8","alias_value":"EIUNGFN6","created_at":"2026-07-05T10:38:03Z"}],"graph_snapshots":[{"event_id":"sha256:8b56ff97e4b9c2e1aeff609035882bf8576f286e251ecabd5bc6f1b2da26ce95","target":"graph","created_at":"2026-07-05T10:38:03Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2503.18130/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement learning from human feedback (RLHF) is an effective method for aligning large language models (LLMs) with human values. However, reward over-optimization remains an open challenge leading to discrepancies between the performance of LLMs under the reward model and the true human objectives. A primary contributor to reward over-optimization is the extrapolation error that arises when the reward model evaluates out-of-distribution (OOD) responses. However, current methods still fail to prevent the increasing frequency of OOD response generation during the reinforcement learning (RL)","authors_text":"Gang Pan, Juntao Dai, Qian Zheng, Taiye Chen, Yaodong Yang","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-03-23T16:20:59Z","title":"Mitigating Reward Over-Optimization in RLHF via Behavior-Supported Regularization"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.18130","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:421c71cd2e08f34bd68ba2906991c5d25fa546bd38c9a2cad7aab19038e692f3","target":"record","created_at":"2026-07-05T10:38:03Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"7ea2787c4e7da787feb91acea81b32f5c5264898cdefbff30c0ec13b30474a16","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-03-23T16:20:59Z","title_canon_sha256":"a63c3e94f31705cb0ab702c2bd7b63449edcd63dc40f5f703f618cda8f372b34"},"schema_version":"1.0","source":{"id":"2503.18130","kind":"arxiv","version":1}},"canonical_sha256":"2228d315bedd3fdf49d0aa07e17c08f5e0310c70c2d47d886e29cac8498f4544","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2228d315bedd3fdf49d0aa07e17c08f5e0310c70c2d47d886e29cac8498f4544","first_computed_at":"2026-07-05T10:38:03.414122Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:38:03.414122Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"JMQu3xBkPghkN1BwQQMcKM39poIL6VowuiEIwJGOkKIb4kamswjTfujwXMBvo8PaRAO73z+fu5TlKURFFrBJCw==","signature_status":"signed_v1","signed_at":"2026-07-05T10:38:03.414569Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.18130","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:421c71cd2e08f34bd68ba2906991c5d25fa546bd38c9a2cad7aab19038e692f3","sha256:8b56ff97e4b9c2e1aeff609035882bf8576f286e251ecabd5bc6f1b2da26ce95"],"state_sha256":"d83023307d7c5373f8f1d0312d24d8a86399629c706f59e6ebd4b1d4d6cc0057"}