{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:NKEDGBHRWJPY3P3B4JRDLQVCOS","short_pith_number":"pith:NKEDGBHR","schema_version":"1.0","canonical_sha256":"6a883304f1b25f8dbf61e26235c2a274a3917240cf481e9b8492ec2ca1d3e800","source":{"kind":"arxiv","id":"2403.18802","version":4},"attestation_state":"computed","paper":{"title":"Long-form factuality in large language models","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Chengrun Yang, Cosmo Du, Da Huang, Daiyi Peng, Dustin Tran, Jerry Wei, Jie Huang, Nathan Hu, Quoc V. Le, Ruibo Liu, Xinying Song, Yifeng Lu","submitted_at":"2024-03-27T17:48:55Z","abstract_excerpt":"Large language models (LLMs) often generate content that contains factual errors when responding to fact-seeking prompts on open-ended topics. To benchmark a model's long-form factuality in open domains, we first use GPT-4 to generate LongFact, a prompt set comprising thousands of questions spanning 38 topics. We then propose that LLM agents can be used as automated evaluators for long-form factuality through a method which we call Search-Augmented Factuality Evaluator (SAFE). SAFE utilizes an LLM to break down a long-form response into a set of individual facts and to evaluate the accuracy of"},"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":"2403.18802","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-27T17:48:55Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"2df7728d2d316131840ef94e90b79244d71a25192a6f7201857f7f9779f1e08b","abstract_canon_sha256":"26b85c588d5015457f50d6799d6f8f7d58aba396e95ddc5ee94e96396d529240"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:32:06.500759Z","signature_b64":"Fs3pwEtNJXG9DYknN/u2jJHNH4YdPl/GJgq4E54T/f35MwowJJQ+Qcd2Bedz9wLsK3giJC7b7SDukmXrTjNmBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6a883304f1b25f8dbf61e26235c2a274a3917240cf481e9b8492ec2ca1d3e800","last_reissued_at":"2026-07-05T09:32:06.500217Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:32:06.500217Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Long-form factuality in large language models","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Chengrun Yang, Cosmo Du, Da Huang, Daiyi Peng, Dustin Tran, Jerry Wei, Jie Huang, Nathan Hu, Quoc V. Le, Ruibo Liu, Xinying Song, Yifeng Lu","submitted_at":"2024-03-27T17:48:55Z","abstract_excerpt":"Large language models (LLMs) often generate content that contains factual errors when responding to fact-seeking prompts on open-ended topics. To benchmark a model's long-form factuality in open domains, we first use GPT-4 to generate LongFact, a prompt set comprising thousands of questions spanning 38 topics. We then propose that LLM agents can be used as automated evaluators for long-form factuality through a method which we call Search-Augmented Factuality Evaluator (SAFE). SAFE utilizes an LLM to break down a long-form response into a set of individual facts and to evaluate the accuracy of"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.18802","kind":"arxiv","version":4},"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/2403.18802/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":"2403.18802","created_at":"2026-07-05T09:32:06.500278+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.18802v4","created_at":"2026-07-05T09:32:06.500278+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.18802","created_at":"2026-07-05T09:32:06.500278+00:00"},{"alias_kind":"pith_short_12","alias_value":"NKEDGBHRWJPY","created_at":"2026-07-05T09:32:06.500278+00:00"},{"alias_kind":"pith_short_16","alias_value":"NKEDGBHRWJPY3P3B","created_at":"2026-07-05T09:32:06.500278+00:00"},{"alias_kind":"pith_short_8","alias_value":"NKEDGBHR","created_at":"2026-07-05T09:32:06.500278+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.29920","citing_title":"Can LLM-as-a-Judge Reliably Verify Rubrics in Agentic Scenarios?","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2406.11354","citing_title":"Preserving Knowledge in Large Language Model with Model-Agnostic Self-Decompression","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17487","citing_title":"Answer Only as Precisely as Justified: Calibrated Claim-Level Specificity Control for Agentic Systems","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08549","citing_title":"VerifAI: A Verifiable Open-Source Search Engine for Biomedical Question Answering","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2411.04368","citing_title":"Measuring short-form factuality in large language models","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2604.03141","citing_title":"Beyond Precision: Importance-Aware Recall for Factuality Evaluation in Long-Form LLM Generation","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10442","citing_title":"StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs","ref_index":115,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10442","citing_title":"StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs","ref_index":115,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23588","citing_title":"FinGround: Detecting and Grounding Financial Hallucinations via Atomic Claim Verification","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17487","citing_title":"Answer Only as Precisely as Justified: Calibrated Claim-Level Specificity Control for Agentic Systems","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NKEDGBHRWJPY3P3B4JRDLQVCOS","json":"https://pith.science/pith/NKEDGBHRWJPY3P3B4JRDLQVCOS.json","graph_json":"https://pith.science/api/pith-number/NKEDGBHRWJPY3P3B4JRDLQVCOS/graph.json","events_json":"https://pith.science/api/pith-number/NKEDGBHRWJPY3P3B4JRDLQVCOS/events.json","paper":"https://pith.science/paper/NKEDGBHR"},"agent_actions":{"view_html":"https://pith.science/pith/NKEDGBHRWJPY3P3B4JRDLQVCOS","download_json":"https://pith.science/pith/NKEDGBHRWJPY3P3B4JRDLQVCOS.json","view_paper":"https://pith.science/paper/NKEDGBHR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.18802&json=true","fetch_graph":"https://pith.science/api/pith-number/NKEDGBHRWJPY3P3B4JRDLQVCOS/graph.json","fetch_events":"https://pith.science/api/pith-number/NKEDGBHRWJPY3P3B4JRDLQVCOS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NKEDGBHRWJPY3P3B4JRDLQVCOS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NKEDGBHRWJPY3P3B4JRDLQVCOS/action/storage_attestation","attest_author":"https://pith.science/pith/NKEDGBHRWJPY3P3B4JRDLQVCOS/action/author_attestation","sign_citation":"https://pith.science/pith/NKEDGBHRWJPY3P3B4JRDLQVCOS/action/citation_signature","submit_replication":"https://pith.science/pith/NKEDGBHRWJPY3P3B4JRDLQVCOS/action/replication_record"}},"created_at":"2026-07-05T09:32:06.500278+00:00","updated_at":"2026-07-05T09:32:06.500278+00:00"}