{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:W5AW73QHXHRS4ZAC566MZ6OMRR","short_pith_number":"pith:W5AW73QH","schema_version":"1.0","canonical_sha256":"b7416fee07b9e32e6402efbcccf9cc8c75905c39fdbe93293cc1ef5a6e1101d5","source":{"kind":"arxiv","id":"2203.11147","version":1},"attestation_state":"computed","paper":{"title":"Teaching language models to support answers with verified quotes","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A 280 billion parameter model can be trained to answer questions with specific cited evidence from documents and to abstain when uncertain.","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Francis Song, Geoffrey Irving, Jacob Menick, John Aslanides, Lucy Campbell-Gillingham, Maja Trebacz, Martin Chadwick, Mia Glaese, Nat McAleese, Susannah Young, Vladimir Mikulik","submitted_at":"2022-03-21T17:26:29Z","abstract_excerpt":"Recent large language models often answer factual questions correctly. But users can't trust any given claim a model makes without fact-checking, because language models can hallucinate convincing nonsense. In this work we use reinforcement learning from human preferences (RLHP) to train \"open-book\" QA models that generate answers whilst also citing specific evidence for their claims, which aids in the appraisal of correctness. Supporting evidence is drawn from multiple documents found via a search engine, or from a single user-provided document. Our 280 billion parameter model, GopherCite, is"},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2203.11147","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-21T17:26:29Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"13c59520008b99b844030d7803e45e1caa24dcb0fc66e41169eb41631e273ec5","abstract_canon_sha256":"3f27c4af1e562a931e61026b868ea299ed8a7336b9e159d0dbe738400d973823"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:14.206967Z","signature_b64":"op1DX+t2vqDDhyd+MoJdrNCcCF3FyRpCPZg7gOtjTgLWJfZofjprS7EdTMsJjC0n6Mz58Vr4mSqN0DpOKl2tCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b7416fee07b9e32e6402efbcccf9cc8c75905c39fdbe93293cc1ef5a6e1101d5","last_reissued_at":"2026-05-17T23:38:14.206266Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:14.206266Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Teaching language models to support answers with verified quotes","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A 280 billion parameter model can be trained to answer questions with specific cited evidence from documents and to abstain when uncertain.","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Francis Song, Geoffrey Irving, Jacob Menick, John Aslanides, Lucy Campbell-Gillingham, Maja Trebacz, Martin Chadwick, Mia Glaese, Nat McAleese, Susannah Young, Vladimir Mikulik","submitted_at":"2022-03-21T17:26:29Z","abstract_excerpt":"Recent large language models often answer factual questions correctly. But users can't trust any given claim a model makes without fact-checking, because language models can hallucinate convincing nonsense. In this work we use reinforcement learning from human preferences (RLHP) to train \"open-book\" QA models that generate answers whilst also citing specific evidence for their claims, which aids in the appraisal of correctness. Supporting evidence is drawn from multiple documents found via a search engine, or from a single user-provided document. Our 280 billion parameter model, GopherCite, is"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our 280 billion parameter model, GopherCite, is able to produce answers with high quality supporting evidence and abstain from answering when unsure. The model's response is found to be high-quality 80% of the time on this Natural Questions subset, and 67% of the time on the ELI5 subset. Abstaining from the third of questions for which it is most unsure improves performance to 90% and 80% respectively.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That human raters' preferences for 'high quality supporting evidence' during RLHP training generalize to produce reliable citations and that the model's internal uncertainty signal for abstention is well-calibrated without introducing new biases.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"GopherCite produces answers with supporting evidence citations, rated high-quality 80% of the time on Natural Questions and 67% on ELI5, improving to 90% and 80% with abstention on uncertain questions.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A 280 billion parameter model can be trained to answer questions with specific cited evidence from documents and to abstain when uncertain.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"0bfbea845d9b4ea5c26a8c4ad7ed798d6d75e5d37e0a6c3c78e702e9d62020fd"},"source":{"id":"2203.11147","kind":"arxiv","version":1},"verdict":{"id":"9612394c-192c-4c1f-a9f9-d2f981f7f4f1","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T11:41:27.964902Z","strongest_claim":"Our 280 billion parameter model, GopherCite, is able to produce answers with high quality supporting evidence and abstain from answering when unsure. The model's response is found to be high-quality 80% of the time on this Natural Questions subset, and 67% of the time on the ELI5 subset. Abstaining from the third of questions for which it is most unsure improves performance to 90% and 80% respectively.","one_line_summary":"GopherCite produces answers with supporting evidence citations, rated high-quality 80% of the time on Natural Questions and 67% on ELI5, improving to 90% and 80% with abstention on uncertain questions.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That human raters' preferences for 'high quality supporting evidence' during RLHP training generalize to produce reliable citations and that the model's internal uncertainty signal for abstention is well-calibrated without introducing new biases.","pith_extraction_headline":"A 280 billion parameter model can be trained to answer questions with specific cited evidence from documents and to abstain when uncertain."},"references":{"count":14,"sample":[{"doi":"10.1145/3041021.3053375","year":2018,"title":"ISBN 9781450349147","work_id":"f9659ba7-d3b9-47d1-a957-64fed829351f","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1177/0894439317715434","year":2018,"title":"road draft tube","work_id":"be985a9d-c962-4bf8-91fa-284ca394a293","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"{url} • {claim} See this fragment from \"{title}\"[1]: {quote}","work_id":"5aad03f9-dc2e-4995-b1ce-97c186899a82","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"{quote}\" Source:","work_id":"23284550-9063-4b4d-90d9-825a613e8ccd","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"What happens if you smash a mirror?","work_id":"712292c7-2504-4989-b456-03e962093e6c","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":14,"snapshot_sha256":"3aa19ac14cbd371c1e693ce659656e65f10ba455726feb7f80c2706295bbbc85","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":"2203.11147","created_at":"2026-05-17T23:38:14.206388+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.11147v1","created_at":"2026-05-17T23:38:14.206388+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.11147","created_at":"2026-05-17T23:38:14.206388+00:00"},{"alias_kind":"pith_short_12","alias_value":"W5AW73QHXHRS","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_16","alias_value":"W5AW73QHXHRS4ZAC","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_8","alias_value":"W5AW73QH","created_at":"2026-05-18T12:33:33.725879+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":30,"internal_anchor_count":30,"sample":[{"citing_arxiv_id":"2211.14275","citing_title":"Solving math word problems with process- and outcome-based feedback","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2312.11805","citing_title":"Gemini: A Family of Highly Capable Multimodal Models","ref_index":62,"is_internal_anchor":true},{"citing_arxiv_id":"2504.12501","citing_title":"Reinforcement Learning from Human Feedback","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2505.04588","citing_title":"ZeroSearch: Incentivize the Search Capability of LLMs without Searching","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2605.00505","citing_title":"LLM-Oriented Information Retrieval: A Denoising-First Perspective","ref_index":128,"is_internal_anchor":true},{"citing_arxiv_id":"2307.06435","citing_title":"A Comprehensive Overview of Large Language Models","ref_index":165,"is_internal_anchor":true},{"citing_arxiv_id":"2505.20825","citing_title":"Reinforced Informativeness Optimization for Long-Form Retrieval-Augmented Generation","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2506.04565","citing_title":"From Standalone LLMs to Integrated Intelligence: A Survey of Compound Al Systems","ref_index":126,"is_internal_anchor":true},{"citing_arxiv_id":"2210.10760","citing_title":"Scaling Laws for Reward Model Overoptimization","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2506.17585","citing_title":"Cite Pretrain: Retrieval-Free Knowledge Attribution for Large Language Models","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2506.19977","citing_title":"Context Attribution with Multi-Armed Bandit Optimization","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2309.11495","citing_title":"Chain-of-Verification Reduces Hallucination in Large Language Models","ref_index":122,"is_internal_anchor":true},{"citing_arxiv_id":"2505.04588","citing_title":"ZeroSearch: Incentivize the Search Capability of LLMs without Searching","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2303.17491","citing_title":"Language Models can Solve Computer Tasks","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"2602.10995","citing_title":"A Human-Centric Framework for Data Attribution in Large Language Models","ref_index":126,"is_internal_anchor":true},{"citing_arxiv_id":"2212.03827","citing_title":"Discovering Latent Knowledge in Language Models Without Supervision","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2603.28281","citing_title":"Corruption-robust Offline Multi-agent Reinforcement Learning From Human Feedback","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2205.06175","citing_title":"A Generalist Agent","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"2310.11511","citing_title":"Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection","ref_index":119,"is_internal_anchor":true},{"citing_arxiv_id":"2204.14198","citing_title":"Flamingo: a Visual Language Model for Few-Shot Learning","ref_index":73,"is_internal_anchor":true},{"citing_arxiv_id":"2604.26981","citing_title":"Budget-Constrained Online Retrieval-Augmented Generation: The Chunk-as-a-Service Model","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2604.25707","citing_title":"From Citation Selection to Citation Absorption: A Measurement Framework for Generative Engine Optimization Across AI Search Platforms","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2305.13301","citing_title":"Training Diffusion Models with Reinforcement Learning","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2605.00505","citing_title":"LLM-Oriented Information Retrieval: A Denoising-First Perspective","ref_index":125,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07153","citing_title":"Beyond Reasoning: Reinforcement Learning Unlocks Parametric Knowledge in LLMs","ref_index":25,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/W5AW73QHXHRS4ZAC566MZ6OMRR","json":"https://pith.science/pith/W5AW73QHXHRS4ZAC566MZ6OMRR.json","graph_json":"https://pith.science/api/pith-number/W5AW73QHXHRS4ZAC566MZ6OMRR/graph.json","events_json":"https://pith.science/api/pith-number/W5AW73QHXHRS4ZAC566MZ6OMRR/events.json","paper":"https://pith.science/paper/W5AW73QH"},"agent_actions":{"view_html":"https://pith.science/pith/W5AW73QHXHRS4ZAC566MZ6OMRR","download_json":"https://pith.science/pith/W5AW73QHXHRS4ZAC566MZ6OMRR.json","view_paper":"https://pith.science/paper/W5AW73QH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.11147&json=true","fetch_graph":"https://pith.science/api/pith-number/W5AW73QHXHRS4ZAC566MZ6OMRR/graph.json","fetch_events":"https://pith.science/api/pith-number/W5AW73QHXHRS4ZAC566MZ6OMRR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W5AW73QHXHRS4ZAC566MZ6OMRR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W5AW73QHXHRS4ZAC566MZ6OMRR/action/storage_attestation","attest_author":"https://pith.science/pith/W5AW73QHXHRS4ZAC566MZ6OMRR/action/author_attestation","sign_citation":"https://pith.science/pith/W5AW73QHXHRS4ZAC566MZ6OMRR/action/citation_signature","submit_replication":"https://pith.science/pith/W5AW73QHXHRS4ZAC566MZ6OMRR/action/replication_record"}},"created_at":"2026-05-17T23:38:14.206388+00:00","updated_at":"2026-05-17T23:38:14.206388+00:00"}