{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:4SA6VYCIUAA6EWBBYCZUSVB5LR","short_pith_number":"pith:4SA6VYCI","schema_version":"1.0","canonical_sha256":"e481eae048a001e25821c0b349543d5c72a42cf44ecb12453e528d2507adb3ab","source":{"kind":"arxiv","id":"2204.07931","version":1},"attestation_state":"computed","paper":{"title":"On the Origin of Hallucinations in Conversational Models: Is it the Datasets or the Models?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Mo Yu, Nouha Dziri, Osmar Zaiane, Sivan Milton, Siva Reddy","submitted_at":"2022-04-17T05:15:24Z","abstract_excerpt":"Knowledge-grounded conversational models are known to suffer from producing factually invalid statements, a phenomenon commonly called hallucination. In this work, we investigate the underlying causes of this phenomenon: is hallucination due to the training data, or to the models? We conduct a comprehensive human study on both existing knowledge-grounded conversational benchmarks and several state-of-the-art models. Our study reveals that the standard benchmarks consist of >60% hallucinated responses, leading to models that not only hallucinate but even amplify hallucinations. Our findings rai"},"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":"2204.07931","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-04-17T05:15:24Z","cross_cats_sorted":[],"title_canon_sha256":"6d75eaa37d76eefe233a3d29de1eab0be00efe1fa7b48ae31070c9ee745c107f","abstract_canon_sha256":"ae35390d5d19ee3c2389a01b7bc8b08e3fc6f317c10551b162ca540e581551bb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:15:19.929551Z","signature_b64":"9RODOW9pYaaiC9C3UtLP/RkLw26Ugw0Ms7kzwZnAcx66yAEjYdwAbuoAy9Y9djUHIstz178SZfun59chwueOCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e481eae048a001e25821c0b349543d5c72a42cf44ecb12453e528d2507adb3ab","last_reissued_at":"2026-07-05T04:15:19.929137Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:15:19.929137Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Origin of Hallucinations in Conversational Models: Is it the Datasets or the Models?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Mo Yu, Nouha Dziri, Osmar Zaiane, Sivan Milton, Siva Reddy","submitted_at":"2022-04-17T05:15:24Z","abstract_excerpt":"Knowledge-grounded conversational models are known to suffer from producing factually invalid statements, a phenomenon commonly called hallucination. In this work, we investigate the underlying causes of this phenomenon: is hallucination due to the training data, or to the models? We conduct a comprehensive human study on both existing knowledge-grounded conversational benchmarks and several state-of-the-art models. Our study reveals that the standard benchmarks consist of >60% hallucinated responses, leading to models that not only hallucinate but even amplify hallucinations. Our findings rai"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2204.07931","kind":"arxiv","version":1},"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/2204.07931/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":"2204.07931","created_at":"2026-07-05T04:15:19.929196+00:00"},{"alias_kind":"arxiv_version","alias_value":"2204.07931v1","created_at":"2026-07-05T04:15:19.929196+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2204.07931","created_at":"2026-07-05T04:15:19.929196+00:00"},{"alias_kind":"pith_short_12","alias_value":"4SA6VYCIUAA6","created_at":"2026-07-05T04:15:19.929196+00:00"},{"alias_kind":"pith_short_16","alias_value":"4SA6VYCIUAA6EWBB","created_at":"2026-07-05T04:15:19.929196+00:00"},{"alias_kind":"pith_short_8","alias_value":"4SA6VYCI","created_at":"2026-07-05T04:15:19.929196+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01612","citing_title":"Scaling with Confidence: Calibrating Confidence of LLMs for Adaptive Test Time Scaling","ref_index":26,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4SA6VYCIUAA6EWBBYCZUSVB5LR","json":"https://pith.science/pith/4SA6VYCIUAA6EWBBYCZUSVB5LR.json","graph_json":"https://pith.science/api/pith-number/4SA6VYCIUAA6EWBBYCZUSVB5LR/graph.json","events_json":"https://pith.science/api/pith-number/4SA6VYCIUAA6EWBBYCZUSVB5LR/events.json","paper":"https://pith.science/paper/4SA6VYCI"},"agent_actions":{"view_html":"https://pith.science/pith/4SA6VYCIUAA6EWBBYCZUSVB5LR","download_json":"https://pith.science/pith/4SA6VYCIUAA6EWBBYCZUSVB5LR.json","view_paper":"https://pith.science/paper/4SA6VYCI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2204.07931&json=true","fetch_graph":"https://pith.science/api/pith-number/4SA6VYCIUAA6EWBBYCZUSVB5LR/graph.json","fetch_events":"https://pith.science/api/pith-number/4SA6VYCIUAA6EWBBYCZUSVB5LR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4SA6VYCIUAA6EWBBYCZUSVB5LR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4SA6VYCIUAA6EWBBYCZUSVB5LR/action/storage_attestation","attest_author":"https://pith.science/pith/4SA6VYCIUAA6EWBBYCZUSVB5LR/action/author_attestation","sign_citation":"https://pith.science/pith/4SA6VYCIUAA6EWBBYCZUSVB5LR/action/citation_signature","submit_replication":"https://pith.science/pith/4SA6VYCIUAA6EWBBYCZUSVB5LR/action/replication_record"}},"created_at":"2026-07-05T04:15:19.929196+00:00","updated_at":"2026-07-05T04:15:19.929196+00:00"}