{"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"}