{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:UHWNDXRKIV2KJB7VQXJZ6TFDA4","short_pith_number":"pith:UHWNDXRK","canonical_record":{"source":{"id":"2302.08582","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-16T21:03:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"28e35a27ebbad37a4a614ae4ffad1b0220df2c1dc09e1119121150333b6adbf3","abstract_canon_sha256":"18e5340257b495e0e163f7db224c4f465647e276f4cc76ef2c42fbd6280abe79"},"schema_version":"1.0"},"canonical_sha256":"a1ecd1de2a4574a487f585d39f4ca30731e689a8e8d438117c743e52b5b85aa5","source":{"kind":"arxiv","id":"2302.08582","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2302.08582","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"arxiv_version","alias_value":"2302.08582v2","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.08582","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"pith_short_12","alias_value":"UHWNDXRKIV2K","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"pith_short_16","alias_value":"UHWNDXRKIV2KJB7V","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"pith_short_8","alias_value":"UHWNDXRK","created_at":"2026-07-05T06:20:40Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:UHWNDXRKIV2KJB7VQXJZ6TFDA4","target":"record","payload":{"canonical_record":{"source":{"id":"2302.08582","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-16T21:03:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"28e35a27ebbad37a4a614ae4ffad1b0220df2c1dc09e1119121150333b6adbf3","abstract_canon_sha256":"18e5340257b495e0e163f7db224c4f465647e276f4cc76ef2c42fbd6280abe79"},"schema_version":"1.0"},"canonical_sha256":"a1ecd1de2a4574a487f585d39f4ca30731e689a8e8d438117c743e52b5b85aa5","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:20:40.807049Z","signature_b64":"T6VRwEIv6HKJYW38A6kak2sugDVL7ri/z5akeF63yyTiNimHAxZi6p3AxJwHdF4HkLIJIwG0oHX31SRHkCToDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a1ecd1de2a4574a487f585d39f4ca30731e689a8e8d438117c743e52b5b85aa5","last_reissued_at":"2026-07-05T06:20:40.806580Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:20:40.806580Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2302.08582","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":"2026-07-05T06:20:40Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"X+FS3IQ32yaN+JIGAxxE2p3Y0zzJXp6wge5WywmZqzrLLv+vzyPgpb6FYfGMlNWxWjDXdt51sWhDp2UdQoFnCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T19:56:57.228645Z"},"content_sha256":"7012e35826160a22d71e82d62c08d87edfbd74e42471d451b9f240a2a1887369","schema_version":"1.0","event_id":"sha256:7012e35826160a22d71e82d62c08d87edfbd74e42471d451b9f240a2a1887369"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:UHWNDXRKIV2KJB7VQXJZ6TFDA4","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Pretraining Language Models with Human Preferences","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Angelica Chen, Christopher L. Buckley, Ethan Perez, Jason Phang, Kejian Shi, Rasika Bhalerao, Samuel R. Bowman, Tomasz Korbak","submitted_at":"2023-02-16T21:03:33Z","abstract_excerpt":"Language models (LMs) are pretrained to imitate internet text, including content that would violate human preferences if generated by an LM: falsehoods, offensive comments, personally identifiable information, low-quality or buggy code, and more. Here, we explore alternative objectives for pretraining LMs in a way that also guides them to generate text aligned with human preferences. We benchmark five objectives for pretraining with human feedback across three tasks and study how they affect the trade-off between alignment and capabilities of pretrained LMs. We find a Pareto-optimal and simple"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.08582","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/2302.08582/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":"2026-07-05T06:20:40Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"P0ve8SjWSvypGBV2OqCVeYilxH38PNlio/29foK9IvCZqb6mDQhmaA45mhZG2V+uw1mk03LgfKuFLjdOcjCbAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T19:56:57.229194Z"},"content_sha256":"dffad27599a2aa7c180656f9c00f0d65864613338e0517d6788205121dfaf9d0","schema_version":"1.0","event_id":"sha256:dffad27599a2aa7c180656f9c00f0d65864613338e0517d6788205121dfaf9d0"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/UHWNDXRKIV2KJB7VQXJZ6TFDA4/bundle.json","state_url":"https://pith.science/pith/UHWNDXRKIV2KJB7VQXJZ6TFDA4/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/UHWNDXRKIV2KJB7VQXJZ6TFDA4/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":"2026-08-12T19:56:57Z","links":{"resolver":"https://pith.science/pith/UHWNDXRKIV2KJB7VQXJZ6TFDA4","bundle":"https://pith.science/pith/UHWNDXRKIV2KJB7VQXJZ6TFDA4/bundle.json","state":"https://pith.science/pith/UHWNDXRKIV2KJB7VQXJZ6TFDA4/state.json","well_known_bundle":"https://pith.science/.well-known/pith/UHWNDXRKIV2KJB7VQXJZ6TFDA4/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:UHWNDXRKIV2KJB7VQXJZ6TFDA4","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":"18e5340257b495e0e163f7db224c4f465647e276f4cc76ef2c42fbd6280abe79","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-16T21:03:33Z","title_canon_sha256":"28e35a27ebbad37a4a614ae4ffad1b0220df2c1dc09e1119121150333b6adbf3"},"schema_version":"1.0","source":{"id":"2302.08582","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2302.08582","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"arxiv_version","alias_value":"2302.08582v2","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.08582","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"pith_short_12","alias_value":"UHWNDXRKIV2K","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"pith_short_16","alias_value":"UHWNDXRKIV2KJB7V","created_at":"2026-07-05T06:20:40Z"},{"alias_kind":"pith_short_8","alias_value":"UHWNDXRK","created_at":"2026-07-05T06:20:40Z"}],"graph_snapshots":[{"event_id":"sha256:dffad27599a2aa7c180656f9c00f0d65864613338e0517d6788205121dfaf9d0","target":"graph","created_at":"2026-07-05T06:20:40Z","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/2302.08582/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Language models (LMs) are pretrained to imitate internet text, including content that would violate human preferences if generated by an LM: falsehoods, offensive comments, personally identifiable information, low-quality or buggy code, and more. Here, we explore alternative objectives for pretraining LMs in a way that also guides them to generate text aligned with human preferences. We benchmark five objectives for pretraining with human feedback across three tasks and study how they affect the trade-off between alignment and capabilities of pretrained LMs. We find a Pareto-optimal and simple","authors_text":"Angelica Chen, Christopher L. Buckley, Ethan Perez, Jason Phang, Kejian Shi, Rasika Bhalerao, Samuel R. Bowman, Tomasz Korbak","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-16T21:03:33Z","title":"Pretraining Language Models with Human Preferences"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.08582","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":"sha256:7012e35826160a22d71e82d62c08d87edfbd74e42471d451b9f240a2a1887369","target":"record","created_at":"2026-07-05T06:20:40Z","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":"18e5340257b495e0e163f7db224c4f465647e276f4cc76ef2c42fbd6280abe79","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-16T21:03:33Z","title_canon_sha256":"28e35a27ebbad37a4a614ae4ffad1b0220df2c1dc09e1119121150333b6adbf3"},"schema_version":"1.0","source":{"id":"2302.08582","kind":"arxiv","version":2}},"canonical_sha256":"a1ecd1de2a4574a487f585d39f4ca30731e689a8e8d438117c743e52b5b85aa5","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a1ecd1de2a4574a487f585d39f4ca30731e689a8e8d438117c743e52b5b85aa5","first_computed_at":"2026-07-05T06:20:40.806580Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T06:20:40.806580Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"T6VRwEIv6HKJYW38A6kak2sugDVL7ri/z5akeF63yyTiNimHAxZi6p3AxJwHdF4HkLIJIwG0oHX31SRHkCToDw==","signature_status":"signed_v1","signed_at":"2026-07-05T06:20:40.807049Z","signed_message":"canonical_sha256_bytes"},"source_id":"2302.08582","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:7012e35826160a22d71e82d62c08d87edfbd74e42471d451b9f240a2a1887369","sha256:dffad27599a2aa7c180656f9c00f0d65864613338e0517d6788205121dfaf9d0"],"state_sha256":"129accbf2d51a445b347ace78d0704e7ab6fee63583fe683ec802bd9a4e30dc3"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"AqCuIrk3JDaWQvKiwMin2MEQ9n+gbDwEATE+Ew9zRCA20P1vV2k1bsJBpLgCqH2uZmb0uuyiY6GIBBAd/RWFDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-12T19:56:57.234224Z","bundle_sha256":"24d3968d142a280a55da6cbb7470ab2abf5dc4f1314e28b5aece1cd153ae370f"}}