{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:NTETKTWBOVTKGAHZWFZVSNI7RJ","short_pith_number":"pith:NTETKTWB","canonical_record":{"source":{"id":"2403.02475","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-04T20:39:24Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"fde3c7c1310928b1e190cacf5a326adbd4d5d68dd78ce1f520c4818bd2e023bc","abstract_canon_sha256":"a550e16cd4ad25dabacb58709a7b9019e26acb1b81bb3b37857a99f7139a9b15"},"schema_version":"1.0"},"canonical_sha256":"6cc9354ec17566a300f9b17359351f8a628aa14d0d09da31060b1c1289498930","source":{"kind":"arxiv","id":"2403.02475","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.02475","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"arxiv_version","alias_value":"2403.02475v1","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.02475","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"pith_short_12","alias_value":"NTETKTWBOVTK","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"pith_short_16","alias_value":"NTETKTWBOVTKGAHZ","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"pith_short_8","alias_value":"NTETKTWB","created_at":"2026-07-05T07:52:18Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:NTETKTWBOVTKGAHZWFZVSNI7RJ","target":"record","payload":{"canonical_record":{"source":{"id":"2403.02475","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-04T20:39:24Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"fde3c7c1310928b1e190cacf5a326adbd4d5d68dd78ce1f520c4818bd2e023bc","abstract_canon_sha256":"a550e16cd4ad25dabacb58709a7b9019e26acb1b81bb3b37857a99f7139a9b15"},"schema_version":"1.0"},"canonical_sha256":"6cc9354ec17566a300f9b17359351f8a628aa14d0d09da31060b1c1289498930","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:52:18.023542Z","signature_b64":"WbtIakVB+5xBmVGnGDewxyiGV4zr1ajJEP9Yr/F7MHi9WWESn7jt1xhdbI+eOVI9oou+yHhEmGDHr9Xm6NhgDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6cc9354ec17566a300f9b17359351f8a628aa14d0d09da31060b1c1289498930","last_reissued_at":"2026-07-05T07:52:18.023064Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:52:18.023064Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2403.02475","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:52:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"SG+Kn1F8YTR12SZWZ/2VU2qGAyhUb76vR0jHWtTDBRAL9j8gwWMDodf0OcSzeabYQOe+DdGpiofzssSY0nFMBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T12:58:46.371748Z"},"content_sha256":"11a756797a0e7b1dfa22a55ce2b983aea88a882242effe0b4e24f9cd02251b4c","schema_version":"1.0","event_id":"sha256:11a756797a0e7b1dfa22a55ce2b983aea88a882242effe0b4e24f9cd02251b4c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:NTETKTWBOVTKGAHZWFZVSNI7RJ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Enhancing LLM Safety via Constrained Direct Preference Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Xiaolin Sun, Zixuan Liu, Zizhan Zheng","submitted_at":"2024-03-04T20:39:24Z","abstract_excerpt":"The rapidly increasing capabilities of large language models (LLMs) raise an urgent need to align AI systems with diverse human preferences to simultaneously enhance their usefulness and safety, despite the often conflicting nature of these goals. To address this important problem, a promising approach is to enforce a safety constraint at the fine-tuning stage through a constrained Reinforcement Learning from Human Feedback (RLHF) framework. This approach, however, is computationally expensive and often unstable. In this work, we introduce Constrained DPO (C-DPO), a novel extension of the rece"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.02475","kind":"arxiv","version":1},"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/2403.02475/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":"2026-07-05T07:52:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"FAL94Z5NrKISdTt3mEcKrj6VYpAMt2lXBg+BM2nRzlV4zeh0JVvERgciefSgnaaIKS6yqsrjXzAaisr0rsmGCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T12:58:46.372339Z"},"content_sha256":"782c441f9f3c57985c566dd9b91f3db6c155d1956f1eb2a39209c208b75216b3","schema_version":"1.0","event_id":"sha256:782c441f9f3c57985c566dd9b91f3db6c155d1956f1eb2a39209c208b75216b3"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/NTETKTWBOVTKGAHZWFZVSNI7RJ/bundle.json","state_url":"https://pith.science/pith/NTETKTWBOVTKGAHZWFZVSNI7RJ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/NTETKTWBOVTKGAHZWFZVSNI7RJ/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":"2026-08-08T12:58:46Z","links":{"resolver":"https://pith.science/pith/NTETKTWBOVTKGAHZWFZVSNI7RJ","bundle":"https://pith.science/pith/NTETKTWBOVTKGAHZWFZVSNI7RJ/bundle.json","state":"https://pith.science/pith/NTETKTWBOVTKGAHZWFZVSNI7RJ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/NTETKTWBOVTKGAHZWFZVSNI7RJ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:NTETKTWBOVTKGAHZWFZVSNI7RJ","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":"a550e16cd4ad25dabacb58709a7b9019e26acb1b81bb3b37857a99f7139a9b15","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-04T20:39:24Z","title_canon_sha256":"fde3c7c1310928b1e190cacf5a326adbd4d5d68dd78ce1f520c4818bd2e023bc"},"schema_version":"1.0","source":{"id":"2403.02475","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.02475","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"arxiv_version","alias_value":"2403.02475v1","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.02475","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"pith_short_12","alias_value":"NTETKTWBOVTK","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"pith_short_16","alias_value":"NTETKTWBOVTKGAHZ","created_at":"2026-07-05T07:52:18Z"},{"alias_kind":"pith_short_8","alias_value":"NTETKTWB","created_at":"2026-07-05T07:52:18Z"}],"graph_snapshots":[{"event_id":"sha256:782c441f9f3c57985c566dd9b91f3db6c155d1956f1eb2a39209c208b75216b3","target":"graph","created_at":"2026-07-05T07:52:18Z","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/2403.02475/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The rapidly increasing capabilities of large language models (LLMs) raise an urgent need to align AI systems with diverse human preferences to simultaneously enhance their usefulness and safety, despite the often conflicting nature of these goals. To address this important problem, a promising approach is to enforce a safety constraint at the fine-tuning stage through a constrained Reinforcement Learning from Human Feedback (RLHF) framework. This approach, however, is computationally expensive and often unstable. In this work, we introduce Constrained DPO (C-DPO), a novel extension of the rece","authors_text":"Xiaolin Sun, Zixuan Liu, Zizhan Zheng","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-04T20:39:24Z","title":"Enhancing LLM Safety via Constrained Direct Preference Optimization"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.02475","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":"sha256:11a756797a0e7b1dfa22a55ce2b983aea88a882242effe0b4e24f9cd02251b4c","target":"record","created_at":"2026-07-05T07:52:18Z","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":"a550e16cd4ad25dabacb58709a7b9019e26acb1b81bb3b37857a99f7139a9b15","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-04T20:39:24Z","title_canon_sha256":"fde3c7c1310928b1e190cacf5a326adbd4d5d68dd78ce1f520c4818bd2e023bc"},"schema_version":"1.0","source":{"id":"2403.02475","kind":"arxiv","version":1}},"canonical_sha256":"6cc9354ec17566a300f9b17359351f8a628aa14d0d09da31060b1c1289498930","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"6cc9354ec17566a300f9b17359351f8a628aa14d0d09da31060b1c1289498930","first_computed_at":"2026-07-05T07:52:18.023064Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:52:18.023064Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"WbtIakVB+5xBmVGnGDewxyiGV4zr1ajJEP9Yr/F7MHi9WWESn7jt1xhdbI+eOVI9oou+yHhEmGDHr9Xm6NhgDw==","signature_status":"signed_v1","signed_at":"2026-07-05T07:52:18.023542Z","signed_message":"canonical_sha256_bytes"},"source_id":"2403.02475","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:11a756797a0e7b1dfa22a55ce2b983aea88a882242effe0b4e24f9cd02251b4c","sha256:782c441f9f3c57985c566dd9b91f3db6c155d1956f1eb2a39209c208b75216b3"],"state_sha256":"01bb6d5e90dd34d686a4d612c31cffb0153f20a7ab119777d096586f162b5a08"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"gZRagUjHYIZN069G7c2Tv1QQ5Rz5eAfhhtmgy+9K22HmhGoJiUX1iEN/YKw/Y1/sEWtcB7UKsPJhWogBqn9LBA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T12:58:46.377219Z","bundle_sha256":"e1b1e89c3bf1ec4a37be967a1f186d9ce054791f2cca6b16ab1e33ee625f9a6f"}}