{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:L72V4I34P6POZ6CZ4QZ6UFMOGW","short_pith_number":"pith:L72V4I34","canonical_record":{"source":{"id":"2404.08495","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-04-12T14:25:49Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"79d9dbeb2f642b0a817b8df4cb7ab1b2ec56e54d840f15c195d5b190e4dd8a48","abstract_canon_sha256":"2d8e0fe7c152ea10edda9d5e82c87427d50b60f0f76fd3b9e22fffc15ead9f52"},"schema_version":"1.0"},"canonical_sha256":"5ff55e237c7f9eecf859e433ea158e358bab18905bc2da16d838cbea48f2d927","source":{"kind":"arxiv","id":"2404.08495","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2404.08495","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"arxiv_version","alias_value":"2404.08495v3","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.08495","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"pith_short_12","alias_value":"L72V4I34P6PO","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"pith_short_16","alias_value":"L72V4I34P6POZ6CZ","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"pith_short_8","alias_value":"L72V4I34","created_at":"2026-07-05T08:08:41Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:L72V4I34P6POZ6CZ4QZ6UFMOGW","target":"record","payload":{"canonical_record":{"source":{"id":"2404.08495","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-04-12T14:25:49Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"79d9dbeb2f642b0a817b8df4cb7ab1b2ec56e54d840f15c195d5b190e4dd8a48","abstract_canon_sha256":"2d8e0fe7c152ea10edda9d5e82c87427d50b60f0f76fd3b9e22fffc15ead9f52"},"schema_version":"1.0"},"canonical_sha256":"5ff55e237c7f9eecf859e433ea158e358bab18905bc2da16d838cbea48f2d927","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:08:41.512409Z","signature_b64":"W5Jp4buahnLZ+baDgMrxSO+ECvnuV26ydgNJURHCvG78Vp2BUpkkKRm8eKGPDgM5r0U8v/ecrDk9myygliNIAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5ff55e237c7f9eecf859e433ea158e358bab18905bc2da16d838cbea48f2d927","last_reissued_at":"2026-07-05T08:08:41.512017Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:08:41.512017Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2404.08495","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:08:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"BQNlXOEj/Dq/nBlFtjZW7mnQNLkCPPL3k3YAKDYjUHn6JXZR4B38AuniruCcrs7+G07Fm/yPlOtBdLk2DQWUDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T03:45:28.991016Z"},"content_sha256":"c3488d4354caa5cf0cc4e21583c7c9131be37aecae967d6feebcfc1df627b569","schema_version":"1.0","event_id":"sha256:c3488d4354caa5cf0cc4e21583c7c9131be37aecae967d6feebcfc1df627b569"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:L72V4I34P6POZ6CZ4QZ6UFMOGW","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Dataset Reset Policy Optimization for RLHF","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Dipendra Misra, Jason D. Lee, Jonathan D. Chang, Kiant\\'e Brantley, Owen Oertell, Wenhao Zhan, Wen Sun","submitted_at":"2024-04-12T14:25:49Z","abstract_excerpt":"Reinforcement Learning (RL) from Human Preference-based feedback is a popular paradigm for fine-tuning generative models, which has produced impressive models such as GPT-4 and Claude3 Opus. This framework often consists of two steps: learning a reward model from an offline preference dataset followed by running online RL to optimize the learned reward model. In this work, leveraging the idea of reset, we propose a new RLHF algorithm with provable guarantees. Motivated by the fact that offline preference dataset provides informative states (i.e., data that is preferred by the labelers), our ne"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.08495","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/2404.08495/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":"2026-07-05T08:08:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Me0BXmPmJ0lHmxPWWRZ32OjVq10TM2azDvTXrwj8xnWRQiAak9uAYA2Gv9uiph0Yr1M4dx560HGOIxsbBwRyDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T03:45:28.991698Z"},"content_sha256":"5d3630d164ad88df98fba6c129eaa460e2e9bead49cb7166e7f8d843a023221c","schema_version":"1.0","event_id":"sha256:5d3630d164ad88df98fba6c129eaa460e2e9bead49cb7166e7f8d843a023221c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/L72V4I34P6POZ6CZ4QZ6UFMOGW/bundle.json","state_url":"https://pith.science/pith/L72V4I34P6POZ6CZ4QZ6UFMOGW/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/L72V4I34P6POZ6CZ4QZ6UFMOGW/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":"2026-08-07T03:45:28Z","links":{"resolver":"https://pith.science/pith/L72V4I34P6POZ6CZ4QZ6UFMOGW","bundle":"https://pith.science/pith/L72V4I34P6POZ6CZ4QZ6UFMOGW/bundle.json","state":"https://pith.science/pith/L72V4I34P6POZ6CZ4QZ6UFMOGW/state.json","well_known_bundle":"https://pith.science/.well-known/pith/L72V4I34P6POZ6CZ4QZ6UFMOGW/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:L72V4I34P6POZ6CZ4QZ6UFMOGW","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":"2d8e0fe7c152ea10edda9d5e82c87427d50b60f0f76fd3b9e22fffc15ead9f52","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-04-12T14:25:49Z","title_canon_sha256":"79d9dbeb2f642b0a817b8df4cb7ab1b2ec56e54d840f15c195d5b190e4dd8a48"},"schema_version":"1.0","source":{"id":"2404.08495","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2404.08495","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"arxiv_version","alias_value":"2404.08495v3","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.08495","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"pith_short_12","alias_value":"L72V4I34P6PO","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"pith_short_16","alias_value":"L72V4I34P6POZ6CZ","created_at":"2026-07-05T08:08:41Z"},{"alias_kind":"pith_short_8","alias_value":"L72V4I34","created_at":"2026-07-05T08:08:41Z"}],"graph_snapshots":[{"event_id":"sha256:5d3630d164ad88df98fba6c129eaa460e2e9bead49cb7166e7f8d843a023221c","target":"graph","created_at":"2026-07-05T08:08:41Z","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/2404.08495/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement Learning (RL) from Human Preference-based feedback is a popular paradigm for fine-tuning generative models, which has produced impressive models such as GPT-4 and Claude3 Opus. This framework often consists of two steps: learning a reward model from an offline preference dataset followed by running online RL to optimize the learned reward model. In this work, leveraging the idea of reset, we propose a new RLHF algorithm with provable guarantees. Motivated by the fact that offline preference dataset provides informative states (i.e., data that is preferred by the labelers), our ne","authors_text":"Dipendra Misra, Jason D. Lee, Jonathan D. Chang, Kiant\\'e Brantley, Owen Oertell, Wenhao Zhan, Wen Sun","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-04-12T14:25:49Z","title":"Dataset Reset Policy Optimization for RLHF"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.08495","kind":"arxiv","version":3},"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:c3488d4354caa5cf0cc4e21583c7c9131be37aecae967d6feebcfc1df627b569","target":"record","created_at":"2026-07-05T08:08:41Z","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":"2d8e0fe7c152ea10edda9d5e82c87427d50b60f0f76fd3b9e22fffc15ead9f52","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-04-12T14:25:49Z","title_canon_sha256":"79d9dbeb2f642b0a817b8df4cb7ab1b2ec56e54d840f15c195d5b190e4dd8a48"},"schema_version":"1.0","source":{"id":"2404.08495","kind":"arxiv","version":3}},"canonical_sha256":"5ff55e237c7f9eecf859e433ea158e358bab18905bc2da16d838cbea48f2d927","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"5ff55e237c7f9eecf859e433ea158e358bab18905bc2da16d838cbea48f2d927","first_computed_at":"2026-07-05T08:08:41.512017Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:08:41.512017Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"W5Jp4buahnLZ+baDgMrxSO+ECvnuV26ydgNJURHCvG78Vp2BUpkkKRm8eKGPDgM5r0U8v/ecrDk9myygliNIAw==","signature_status":"signed_v1","signed_at":"2026-07-05T08:08:41.512409Z","signed_message":"canonical_sha256_bytes"},"source_id":"2404.08495","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c3488d4354caa5cf0cc4e21583c7c9131be37aecae967d6feebcfc1df627b569","sha256:5d3630d164ad88df98fba6c129eaa460e2e9bead49cb7166e7f8d843a023221c"],"state_sha256":"8a5be922cd57e27c452f8d7c333a2b0bdd80e458d4d321fdfcd4d12c67eae0b1"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Es1mR5TfUQwsHjZfFyC/jg5YknyEQk0RK0PLc/cXFDMPTHxJg7TtbR7+UXC9QqZji2ESOmpcZo9bRw64gb+nCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T03:45:28.995480Z","bundle_sha256":"be4c985423510336a75590cebe01923c4546f96882b265dadd61e709df2f8fa1"}}