{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:6RCTS72FXC5VNECUJ6VNNWEYIF","short_pith_number":"pith:6RCTS72F","schema_version":"1.0","canonical_sha256":"f445397f45b8bb5690544faad6d898414c2ae56e4ddc8c4e0e4b62136fa139de","source":{"kind":"arxiv","id":"2401.12292","version":2},"attestation_state":"computed","paper":{"title":"GRATH: Gradual Self-Truthifying for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bo Li, Dawn Song, Weixin Chen","submitted_at":"2024-01-22T19:00:08Z","abstract_excerpt":"Truthfulness is paramount for large language models (LLMs) as they are increasingly deployed in real-world applications. However, existing LLMs still struggle with generating truthful content, as evidenced by their modest performance on benchmarks like TruthfulQA. To address this issue, we propose GRAdual self-truTHifying (GRATH), a novel post-processing method to enhance truthfulness of LLMs. GRATH utilizes out-of-domain question prompts to generate pairwise truthfulness training data with each pair containing a question and its correct and incorrect answers, and then optimizes the model via "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2401.12292","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-22T19:00:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"35ca42f3f92347d2844899666a176f79f68889edbbc656941e0281de26176f81","abstract_canon_sha256":"3a3f398b49cafa1ed7e56989fb39544ee73fee089d9a399482982c7840299602"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:39:34.094543Z","signature_b64":"Tu3LKZvsK63ewoav29uhYVMFmuK0vLksdZ7O+yi15YbdjzA1yEh+XuRrFX/KRzgqY6OAjgAzQ1dBBG4pczhmAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f445397f45b8bb5690544faad6d898414c2ae56e4ddc8c4e0e4b62136fa139de","last_reissued_at":"2026-07-05T07:39:34.094100Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:39:34.094100Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GRATH: Gradual Self-Truthifying for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bo Li, Dawn Song, Weixin Chen","submitted_at":"2024-01-22T19:00:08Z","abstract_excerpt":"Truthfulness is paramount for large language models (LLMs) as they are increasingly deployed in real-world applications. However, existing LLMs still struggle with generating truthful content, as evidenced by their modest performance on benchmarks like TruthfulQA. To address this issue, we propose GRAdual self-truTHifying (GRATH), a novel post-processing method to enhance truthfulness of LLMs. GRATH utilizes out-of-domain question prompts to generate pairwise truthfulness training data with each pair containing a question and its correct and incorrect answers, and then optimizes the model via "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.12292","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/2401.12292/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2401.12292","created_at":"2026-07-05T07:39:34.094156+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.12292v2","created_at":"2026-07-05T07:39:34.094156+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.12292","created_at":"2026-07-05T07:39:34.094156+00:00"},{"alias_kind":"pith_short_12","alias_value":"6RCTS72FXC5V","created_at":"2026-07-05T07:39:34.094156+00:00"},{"alias_kind":"pith_short_16","alias_value":"6RCTS72FXC5VNECU","created_at":"2026-07-05T07:39:34.094156+00:00"},{"alias_kind":"pith_short_8","alias_value":"6RCTS72F","created_at":"2026-07-05T07:39:34.094156+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2403.07691","citing_title":"ORPO: Monolithic Preference Optimization without Reference Model","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6RCTS72FXC5VNECUJ6VNNWEYIF","json":"https://pith.science/pith/6RCTS72FXC5VNECUJ6VNNWEYIF.json","graph_json":"https://pith.science/api/pith-number/6RCTS72FXC5VNECUJ6VNNWEYIF/graph.json","events_json":"https://pith.science/api/pith-number/6RCTS72FXC5VNECUJ6VNNWEYIF/events.json","paper":"https://pith.science/paper/6RCTS72F"},"agent_actions":{"view_html":"https://pith.science/pith/6RCTS72FXC5VNECUJ6VNNWEYIF","download_json":"https://pith.science/pith/6RCTS72FXC5VNECUJ6VNNWEYIF.json","view_paper":"https://pith.science/paper/6RCTS72F","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.12292&json=true","fetch_graph":"https://pith.science/api/pith-number/6RCTS72FXC5VNECUJ6VNNWEYIF/graph.json","fetch_events":"https://pith.science/api/pith-number/6RCTS72FXC5VNECUJ6VNNWEYIF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6RCTS72FXC5VNECUJ6VNNWEYIF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6RCTS72FXC5VNECUJ6VNNWEYIF/action/storage_attestation","attest_author":"https://pith.science/pith/6RCTS72FXC5VNECUJ6VNNWEYIF/action/author_attestation","sign_citation":"https://pith.science/pith/6RCTS72FXC5VNECUJ6VNNWEYIF/action/citation_signature","submit_replication":"https://pith.science/pith/6RCTS72FXC5VNECUJ6VNNWEYIF/action/replication_record"}},"created_at":"2026-07-05T07:39:34.094156+00:00","updated_at":"2026-07-05T07:39:34.094156+00:00"}