{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:BAPKPM2A7PECQXIUJ2GNDZ3ILG","short_pith_number":"pith:BAPKPM2A","canonical_record":{"source":{"id":"2406.01013","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-06-03T05:46:53Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"d74ff69db6e93de39b571e17fa53610dedafa980d90cf5628bb600d0cf0fb5f8","abstract_canon_sha256":"671bd616d612963c99f15394ca187ce16429a5b3842c8accfe57b3c3aeee3565"},"schema_version":"1.0"},"canonical_sha256":"081ea7b340fbc8285d144e8cd1e76859a3df7156ccf9fd59c8abc886c6fc39db","source":{"kind":"arxiv","id":"2406.01013","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.01013","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"arxiv_version","alias_value":"2406.01013v2","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.01013","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"pith_short_12","alias_value":"BAPKPM2A7PEC","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"pith_short_16","alias_value":"BAPKPM2A7PECQXIU","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"pith_short_8","alias_value":"BAPKPM2A","created_at":"2026-07-05T08:34:19Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:BAPKPM2A7PECQXIUJ2GNDZ3ILG","target":"record","payload":{"canonical_record":{"source":{"id":"2406.01013","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-06-03T05:46:53Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"d74ff69db6e93de39b571e17fa53610dedafa980d90cf5628bb600d0cf0fb5f8","abstract_canon_sha256":"671bd616d612963c99f15394ca187ce16429a5b3842c8accfe57b3c3aeee3565"},"schema_version":"1.0"},"canonical_sha256":"081ea7b340fbc8285d144e8cd1e76859a3df7156ccf9fd59c8abc886c6fc39db","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:34:19.966174Z","signature_b64":"k2vBDLnkxHij6SM1BDWHEsR1TuMEh3PhtzpXeN0WsJViDkrebYVkrXgSFcygP62QbQzKMpoEcACCANTfmKynBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"081ea7b340fbc8285d144e8cd1e76859a3df7156ccf9fd59c8abc886c6fc39db","last_reissued_at":"2026-07-05T08:34:19.965720Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:34:19.965720Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2406.01013","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:34:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1L+ZfMVYWcSc4zqFTsFRToW08Jd9zCgeaAmMv+r2044ilPLLGvLOZ+O/sxdNV9ldP6Q6Uprg+EJ4s/NFSKM/DQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T23:33:31.052406Z"},"content_sha256":"fe37b850b363a1c381954d7da178e54dce0caa678e5518b97c0977e03286b331","schema_version":"1.0","event_id":"sha256:fe37b850b363a1c381954d7da178e54dce0caa678e5518b97c0977e03286b331"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:BAPKPM2A7PECQXIUJ2GNDZ3ILG","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Scalable Ensembling For Mitigating Reward Overoptimisation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Ahmed M. Ahmed, Rafael Rafailov, Sanmi Koyejo, Stepan Sharkov, Xuechen Li","submitted_at":"2024-06-03T05:46:53Z","abstract_excerpt":"Reinforcement Learning from Human Feedback (RLHF) has enabled significant advancements within language modeling for powerful, instruction-following models. However, the alignment of these models remains a pressing challenge as the policy tends to overfit the learned ``proxy\" reward model past an inflection point of utility as measured by a ``gold\" reward model that is more performant -- a phenomenon known as overoptimisation. Prior work has mitigated this issue by computing a pessimistic statistic over an ensemble of reward models, which is common in Offline Reinforcement Learning but incredib"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.01013","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/2406.01013/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:34:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"v6hWrLMiI7YHSXHNMD/ZKB7DWfZKuBSWlp/r2D2bCE66YHrcvDAY1ypHBczX/3Oc5vPk9RynjueGY2FUQgduDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T23:33:31.052913Z"},"content_sha256":"e7be50246cb3e924e59bb1045df1c5d85d730fdbb69959a5c6c1cc2a79694437","schema_version":"1.0","event_id":"sha256:e7be50246cb3e924e59bb1045df1c5d85d730fdbb69959a5c6c1cc2a79694437"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/BAPKPM2A7PECQXIUJ2GNDZ3ILG/bundle.json","state_url":"https://pith.science/pith/BAPKPM2A7PECQXIUJ2GNDZ3ILG/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/BAPKPM2A7PECQXIUJ2GNDZ3ILG/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-08T23:33:31Z","links":{"resolver":"https://pith.science/pith/BAPKPM2A7PECQXIUJ2GNDZ3ILG","bundle":"https://pith.science/pith/BAPKPM2A7PECQXIUJ2GNDZ3ILG/bundle.json","state":"https://pith.science/pith/BAPKPM2A7PECQXIUJ2GNDZ3ILG/state.json","well_known_bundle":"https://pith.science/.well-known/pith/BAPKPM2A7PECQXIUJ2GNDZ3ILG/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:BAPKPM2A7PECQXIUJ2GNDZ3ILG","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":"671bd616d612963c99f15394ca187ce16429a5b3842c8accfe57b3c3aeee3565","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-06-03T05:46:53Z","title_canon_sha256":"d74ff69db6e93de39b571e17fa53610dedafa980d90cf5628bb600d0cf0fb5f8"},"schema_version":"1.0","source":{"id":"2406.01013","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.01013","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"arxiv_version","alias_value":"2406.01013v2","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.01013","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"pith_short_12","alias_value":"BAPKPM2A7PEC","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"pith_short_16","alias_value":"BAPKPM2A7PECQXIU","created_at":"2026-07-05T08:34:19Z"},{"alias_kind":"pith_short_8","alias_value":"BAPKPM2A","created_at":"2026-07-05T08:34:19Z"}],"graph_snapshots":[{"event_id":"sha256:e7be50246cb3e924e59bb1045df1c5d85d730fdbb69959a5c6c1cc2a79694437","target":"graph","created_at":"2026-07-05T08:34:19Z","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/2406.01013/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement Learning from Human Feedback (RLHF) has enabled significant advancements within language modeling for powerful, instruction-following models. However, the alignment of these models remains a pressing challenge as the policy tends to overfit the learned ``proxy\" reward model past an inflection point of utility as measured by a ``gold\" reward model that is more performant -- a phenomenon known as overoptimisation. Prior work has mitigated this issue by computing a pessimistic statistic over an ensemble of reward models, which is common in Offline Reinforcement Learning but incredib","authors_text":"Ahmed M. Ahmed, Rafael Rafailov, Sanmi Koyejo, Stepan Sharkov, Xuechen Li","cross_cats":["cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-06-03T05:46:53Z","title":"Scalable Ensembling For Mitigating Reward Overoptimisation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.01013","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:fe37b850b363a1c381954d7da178e54dce0caa678e5518b97c0977e03286b331","target":"record","created_at":"2026-07-05T08:34:19Z","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":"671bd616d612963c99f15394ca187ce16429a5b3842c8accfe57b3c3aeee3565","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-06-03T05:46:53Z","title_canon_sha256":"d74ff69db6e93de39b571e17fa53610dedafa980d90cf5628bb600d0cf0fb5f8"},"schema_version":"1.0","source":{"id":"2406.01013","kind":"arxiv","version":2}},"canonical_sha256":"081ea7b340fbc8285d144e8cd1e76859a3df7156ccf9fd59c8abc886c6fc39db","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"081ea7b340fbc8285d144e8cd1e76859a3df7156ccf9fd59c8abc886c6fc39db","first_computed_at":"2026-07-05T08:34:19.965720Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:34:19.965720Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"k2vBDLnkxHij6SM1BDWHEsR1TuMEh3PhtzpXeN0WsJViDkrebYVkrXgSFcygP62QbQzKMpoEcACCANTfmKynBQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:34:19.966174Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.01013","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:fe37b850b363a1c381954d7da178e54dce0caa678e5518b97c0977e03286b331","sha256:e7be50246cb3e924e59bb1045df1c5d85d730fdbb69959a5c6c1cc2a79694437"],"state_sha256":"2c354f9f0348f235bbae445bf7374b6eaa2e99cf1267800ce8c6bab430f2ba13"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PG2ist6h4TpdnqluKMMK8e16qWrY7sVoPq8MOG20qtPOgzQn/rQdAGUVqvb7a47Y/ooaPpHrFLZF0thKKEsiDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T23:33:31.057408Z","bundle_sha256":"d428a44cd90e96258dc50555b9409ffecbcf0fdd8962a7189d276f70cfa9f7d4"}}