{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:CA27LYLPCGEFFCW2HWFLZMHWYB","short_pith_number":"pith:CA27LYLP","canonical_record":{"source":{"id":"2407.16637","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-23T16:54:28Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"f4245f96ff66064442d091a2505b5a3ad69f600440462d661edce87f7988555a","abstract_canon_sha256":"59d0268e9542f9123c66502fcd164522800ef65cc3472d3fe5fe48ce2199537e"},"schema_version":"1.0"},"canonical_sha256":"1035f5e16f1188528ada3d8abcb0f6c0714a3db1e764d97dc2efd91605cf8b26","source":{"kind":"arxiv","id":"2407.16637","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.16637","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"arxiv_version","alias_value":"2407.16637v2","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.16637","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"pith_short_12","alias_value":"CA27LYLPCGEF","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"pith_short_16","alias_value":"CA27LYLPCGEFFCW2","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"pith_short_8","alias_value":"CA27LYLP","created_at":"2026-07-05T09:26:29Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:CA27LYLPCGEFFCW2HWFLZMHWYB","target":"record","payload":{"canonical_record":{"source":{"id":"2407.16637","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-23T16:54:28Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"f4245f96ff66064442d091a2505b5a3ad69f600440462d661edce87f7988555a","abstract_canon_sha256":"59d0268e9542f9123c66502fcd164522800ef65cc3472d3fe5fe48ce2199537e"},"schema_version":"1.0"},"canonical_sha256":"1035f5e16f1188528ada3d8abcb0f6c0714a3db1e764d97dc2efd91605cf8b26","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:26:29.319459Z","signature_b64":"lpoOVWj0LVuXEfo2c6iZA3maAy9NqeVSM3VfZ9G79+TbAfJUDDPtjJfq1t5Z5iyYpHCid7lqqJSOMG9gCVVyBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1035f5e16f1188528ada3d8abcb0f6c0714a3db1e764d97dc2efd91605cf8b26","last_reissued_at":"2026-07-05T09:26:29.318968Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:26:29.318968Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2407.16637","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":"2026-07-05T09:26:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"iujy79BA60ctmspa+jR3vodiFq826JaVG03v3JMQaCy/a4g3tqpgSk3rLfIWEQb/Qvm3xlUTR3bG3Dl4DPkTCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T23:33:46.045628Z"},"content_sha256":"102d4fc716959e1abb63d443f315059aa6b3a30eee78b99135386ec7c91059d6","schema_version":"1.0","event_id":"sha256:102d4fc716959e1abb63d443f315059aa6b3a30eee78b99135386ec7c91059d6"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:CA27LYLPCGEFFCW2HWFLZMHWYB","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Course-Correction: Safety Alignment Using Synthetic Preferences","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Haiqin Weng, Han Qiu, Renjie Gu, Rongwu Xu, Tianwei Zhang, Wei Xu, Yan Liu, Yishuo Cai, Zhenhong Zhou","submitted_at":"2024-07-23T16:54:28Z","abstract_excerpt":"The risk of harmful content generated by large language models (LLMs) becomes a critical concern. This paper presents a systematic study on assessing and improving LLMs' capability to perform the task of \\textbf{course-correction}, \\ie, the model can steer away from generating harmful content autonomously. To start with, we introduce the \\textsc{C$^2$-Eval} benchmark for quantitative assessment and analyze 10 popular LLMs, revealing varying proficiency of current safety-tuned LLMs in course-correction. To improve, we propose fine-tuning LLMs with preference learning, emphasizing the preference"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.16637","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/2407.16637/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":"2026-07-05T09:26:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"CLg+sX8XR5bdvl2fVsEA7mGqSSJjx3lQp3PvPNSjN8dQz3D/8jdUpbI/Q5IEWJxcTNnxq3qAAjT4DIZhODctCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T23:33:46.046148Z"},"content_sha256":"40676aad76ee78d7d136525ed4d776c9153e79018aaeb1858dd00b0e5d681d1e","schema_version":"1.0","event_id":"sha256:40676aad76ee78d7d136525ed4d776c9153e79018aaeb1858dd00b0e5d681d1e"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/CA27LYLPCGEFFCW2HWFLZMHWYB/bundle.json","state_url":"https://pith.science/pith/CA27LYLPCGEFFCW2HWFLZMHWYB/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/CA27LYLPCGEFFCW2HWFLZMHWYB/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":"2026-08-08T23:33:46Z","links":{"resolver":"https://pith.science/pith/CA27LYLPCGEFFCW2HWFLZMHWYB","bundle":"https://pith.science/pith/CA27LYLPCGEFFCW2HWFLZMHWYB/bundle.json","state":"https://pith.science/pith/CA27LYLPCGEFFCW2HWFLZMHWYB/state.json","well_known_bundle":"https://pith.science/.well-known/pith/CA27LYLPCGEFFCW2HWFLZMHWYB/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:CA27LYLPCGEFFCW2HWFLZMHWYB","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":"59d0268e9542f9123c66502fcd164522800ef65cc3472d3fe5fe48ce2199537e","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-23T16:54:28Z","title_canon_sha256":"f4245f96ff66064442d091a2505b5a3ad69f600440462d661edce87f7988555a"},"schema_version":"1.0","source":{"id":"2407.16637","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.16637","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"arxiv_version","alias_value":"2407.16637v2","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.16637","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"pith_short_12","alias_value":"CA27LYLPCGEF","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"pith_short_16","alias_value":"CA27LYLPCGEFFCW2","created_at":"2026-07-05T09:26:29Z"},{"alias_kind":"pith_short_8","alias_value":"CA27LYLP","created_at":"2026-07-05T09:26:29Z"}],"graph_snapshots":[{"event_id":"sha256:40676aad76ee78d7d136525ed4d776c9153e79018aaeb1858dd00b0e5d681d1e","target":"graph","created_at":"2026-07-05T09:26:29Z","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/2407.16637/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The risk of harmful content generated by large language models (LLMs) becomes a critical concern. This paper presents a systematic study on assessing and improving LLMs' capability to perform the task of \\textbf{course-correction}, \\ie, the model can steer away from generating harmful content autonomously. To start with, we introduce the \\textsc{C$^2$-Eval} benchmark for quantitative assessment and analyze 10 popular LLMs, revealing varying proficiency of current safety-tuned LLMs in course-correction. To improve, we propose fine-tuning LLMs with preference learning, emphasizing the preference","authors_text":"Haiqin Weng, Han Qiu, Renjie Gu, Rongwu Xu, Tianwei Zhang, Wei Xu, Yan Liu, Yishuo Cai, Zhenhong Zhou","cross_cats":["cs.AI","cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-23T16:54:28Z","title":"Course-Correction: Safety Alignment Using Synthetic Preferences"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.16637","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":"sha256:102d4fc716959e1abb63d443f315059aa6b3a30eee78b99135386ec7c91059d6","target":"record","created_at":"2026-07-05T09:26:29Z","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":"59d0268e9542f9123c66502fcd164522800ef65cc3472d3fe5fe48ce2199537e","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-23T16:54:28Z","title_canon_sha256":"f4245f96ff66064442d091a2505b5a3ad69f600440462d661edce87f7988555a"},"schema_version":"1.0","source":{"id":"2407.16637","kind":"arxiv","version":2}},"canonical_sha256":"1035f5e16f1188528ada3d8abcb0f6c0714a3db1e764d97dc2efd91605cf8b26","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1035f5e16f1188528ada3d8abcb0f6c0714a3db1e764d97dc2efd91605cf8b26","first_computed_at":"2026-07-05T09:26:29.318968Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:26:29.318968Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"lpoOVWj0LVuXEfo2c6iZA3maAy9NqeVSM3VfZ9G79+TbAfJUDDPtjJfq1t5Z5iyYpHCid7lqqJSOMG9gCVVyBQ==","signature_status":"signed_v1","signed_at":"2026-07-05T09:26:29.319459Z","signed_message":"canonical_sha256_bytes"},"source_id":"2407.16637","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:102d4fc716959e1abb63d443f315059aa6b3a30eee78b99135386ec7c91059d6","sha256:40676aad76ee78d7d136525ed4d776c9153e79018aaeb1858dd00b0e5d681d1e"],"state_sha256":"28ffe5232c1021c1eca8b888a77e82ca053812db96db319a196f41fecc3c6f10"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Kz/BiNM/0O3Bz+7vn3FQzQ02FlpejZIABEY+ayRlEBzkapWDwJ6ln5KCwTYjI2dqM/B1s1lU+YIVD6ImufM+Aw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T23:33:46.049581Z","bundle_sha256":"bef7ea8e06544776838c56373012de2409221413b3aece81e513a39afdf7f071"}}