{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:5FWM6ENK2CNBXOSQUVM46TRFTD","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":"2c2e44973de72317414e68684976a097cf151b7a7198018ff27c73f72b26895a","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-20T18:20:59Z","title_canon_sha256":"5ff4f7778d156ed353f1692dc075df8f93cda31b80614d20bd1d2ab4f42adcd6"},"schema_version":"1.0","source":{"id":"2402.13210","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.13210","created_at":"2026-07-05T08:39:31Z"},{"alias_kind":"arxiv_version","alias_value":"2402.13210v2","created_at":"2026-07-05T08:39:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.13210","created_at":"2026-07-05T08:39:31Z"},{"alias_kind":"pith_short_12","alias_value":"5FWM6ENK2CNB","created_at":"2026-07-05T08:39:31Z"},{"alias_kind":"pith_short_16","alias_value":"5FWM6ENK2CNBXOSQ","created_at":"2026-07-05T08:39:31Z"},{"alias_kind":"pith_short_8","alias_value":"5FWM6ENK","created_at":"2026-07-05T08:39:31Z"}],"graph_snapshots":[{"event_id":"sha256:e84eb17b8c465a3399c34d00ae33349d9f830b846af07fd8c98f9014273f8c32","target":"graph","created_at":"2026-07-05T08:39:31Z","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/2402.13210/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"To ensure that large language model (LLM) responses are helpful and non-toxic, a reward model trained on human preference data is usually used. LLM responses with high rewards are then selected through best-of-$n$ (BoN) sampling or the LLM is further optimized to produce responses with high rewards through reinforcement learning from human feedback (RLHF). However, these processes are susceptible to reward overoptimization or `hacking', where responses receive high rewards due to imperfections in the reward model rather than true preference, particularly as prompts or responses deviate from th","authors_text":"Adam X. Yang, Haitham Bou-Ammar, Jun Wang, Laurence Aitchison, Maxime Robeyns, Thomas Coste, Zhengyan Shi","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-20T18:20:59Z","title":"Bayesian Reward Models for LLM Alignment"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.13210","kind":"arxiv","version":2},"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:e935f39e25150fea756d1a614e0a6db3e83c142b65948d65cc5de66de9258794","target":"record","created_at":"2026-07-05T08:39:31Z","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":"2c2e44973de72317414e68684976a097cf151b7a7198018ff27c73f72b26895a","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-20T18:20:59Z","title_canon_sha256":"5ff4f7778d156ed353f1692dc075df8f93cda31b80614d20bd1d2ab4f42adcd6"},"schema_version":"1.0","source":{"id":"2402.13210","kind":"arxiv","version":2}},"canonical_sha256":"e96ccf11aad09a1bba50a559cf4e2598e6902502aee109858379819890b025a3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e96ccf11aad09a1bba50a559cf4e2598e6902502aee109858379819890b025a3","first_computed_at":"2026-07-05T08:39:31.279957Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:39:31.279957Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"PA0qrfUDgsm9t6FQ40OKCMAP+001axEtqy4TR0zU08aCU6nnJC1XZILImbszagF3gjALzgpzP8xx96dcM3+rCQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:39:31.280457Z","signed_message":"canonical_sha256_bytes"},"source_id":"2402.13210","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e935f39e25150fea756d1a614e0a6db3e83c142b65948d65cc5de66de9258794","sha256:e84eb17b8c465a3399c34d00ae33349d9f830b846af07fd8c98f9014273f8c32"],"state_sha256":"f2fe0943d8b1aca20eb4bb62f5823db816a8d127f01926b47e226141f5826dee"}