{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:HISRIZNZQDMZIGTIE5EFOEQ3SB","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":"b85129e422a59c958cc2dedaba21d08198a90bb423d92972ef6a250de4af3917","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-12-12T06:10:42Z","title_canon_sha256":"7d866dab8f182a90948bc12456b3386e18fa8e988567303bed4b548ec5f07d94"},"schema_version":"1.0","source":{"id":"2312.07000","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2312.07000","created_at":"2026-07-05T09:26:57Z"},{"alias_kind":"arxiv_version","alias_value":"2312.07000v2","created_at":"2026-07-05T09:26:57Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.07000","created_at":"2026-07-05T09:26:57Z"},{"alias_kind":"pith_short_12","alias_value":"HISRIZNZQDMZ","created_at":"2026-07-05T09:26:57Z"},{"alias_kind":"pith_short_16","alias_value":"HISRIZNZQDMZIGTI","created_at":"2026-07-05T09:26:57Z"},{"alias_kind":"pith_short_8","alias_value":"HISRIZNZ","created_at":"2026-07-05T09:26:57Z"}],"graph_snapshots":[{"event_id":"sha256:2ace6de9a0773457d2c33a710aee0986abdadba627784b2df66a33bf9483dee4","target":"graph","created_at":"2026-07-05T09:26:57Z","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/2312.07000/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recent research has made significant strides in aligning large language models (LLMs) with helpfulness and harmlessness. In this paper, we argue for the importance of alignment for \\emph{honesty}, ensuring that LLMs proactively refuse to answer questions when they lack knowledge, while still not being overly conservative. However, a pivotal aspect of alignment for honesty involves discerning an LLM's knowledge boundaries, which demands comprehensive solutions in terms of metric development, benchmark creation, and training methodologies. We address these challenges by first establishing a prec","authors_text":"Ethan Chern, Graham Neubig, Pengfei Liu, Xipeng Qiu, Yuqing Yang","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-12-12T06:10:42Z","title":"Alignment for Honesty"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.07000","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:fefdf2817896386e62b982300e80c88d7b4a54f906f3a24c162492f2b7aa1de8","target":"record","created_at":"2026-07-05T09:26:57Z","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":"b85129e422a59c958cc2dedaba21d08198a90bb423d92972ef6a250de4af3917","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-12-12T06:10:42Z","title_canon_sha256":"7d866dab8f182a90948bc12456b3386e18fa8e988567303bed4b548ec5f07d94"},"schema_version":"1.0","source":{"id":"2312.07000","kind":"arxiv","version":2}},"canonical_sha256":"3a251465b980d9941a68274857121b904eb0bdb462a7f9854fb77a8b3f0f100a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3a251465b980d9941a68274857121b904eb0bdb462a7f9854fb77a8b3f0f100a","first_computed_at":"2026-07-05T09:26:57.462016Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:26:57.462016Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"THmM4W/GWI+wviTqu7oCsx2t/u+Jr/WJrljeNjHZplxAGjd0UaBkkSxNrUWyaQy5ZY8unU2DASysDabfh8jYDA==","signature_status":"signed_v1","signed_at":"2026-07-05T09:26:57.462579Z","signed_message":"canonical_sha256_bytes"},"source_id":"2312.07000","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:fefdf2817896386e62b982300e80c88d7b4a54f906f3a24c162492f2b7aa1de8","sha256:2ace6de9a0773457d2c33a710aee0986abdadba627784b2df66a33bf9483dee4"],"state_sha256":"83730874001e9db379cd116cc63370855c9c9497316c368baae19539e9100b64"}