{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:YYLIFSUQRJMM2PUX7F3FOBH56O","short_pith_number":"pith:YYLIFSUQ","schema_version":"1.0","canonical_sha256":"c61682ca908a58cd3e97f9765704fdf3828020d71aef902fda742ded352fa82c","source":{"kind":"arxiv","id":"2404.04298","version":3},"attestation_state":"computed","paper":{"title":"SELF-[IN]CORRECT: LLMs Struggle with Discriminating Self-Generated Responses","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Benjamin Van Durme, Daniel Khashabi, Dongwei Jiang, Jingyu Zhang, Nathaniel Weir, Orion Weller","submitted_at":"2024-04-04T20:27:37Z","abstract_excerpt":"Can LLMs consistently improve their previous outputs for better results? For this to be true, LLMs would need to be better at discriminating among previously-generated alternatives, than generating initial responses. We explore the validity of this hypothesis in practice. We first formulate a unified framework that allows us to compare the generative and discriminative capability of any model on any task. In our resulting experimental analysis of several open-source and industrial LLMs, we observe that models are not reliably better at discriminating among previously-generated alternatives tha"},"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":"2404.04298","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-04-04T20:27:37Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"640933ad3495403f07dfff78c44a2f435e767927f47662a8d7bec371ac8a90ba","abstract_canon_sha256":"a73e75a933ebecd64d30fa7eac4bc6a3f35c2544c99c68e5828a7c6e324c4655"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:03:50.920712Z","signature_b64":"WhkpH0D0nS/zWpjDzGElui5OHBzELS9skOyTHJhKEWddph115++VaObf3BOYYkWjxE2PuayJH0ceHnZg2d5sAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c61682ca908a58cd3e97f9765704fdf3828020d71aef902fda742ded352fa82c","last_reissued_at":"2026-07-05T09:03:50.920172Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:03:50.920172Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SELF-[IN]CORRECT: LLMs Struggle with Discriminating Self-Generated Responses","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Benjamin Van Durme, Daniel Khashabi, Dongwei Jiang, Jingyu Zhang, Nathaniel Weir, Orion Weller","submitted_at":"2024-04-04T20:27:37Z","abstract_excerpt":"Can LLMs consistently improve their previous outputs for better results? For this to be true, LLMs would need to be better at discriminating among previously-generated alternatives, than generating initial responses. We explore the validity of this hypothesis in practice. We first formulate a unified framework that allows us to compare the generative and discriminative capability of any model on any task. In our resulting experimental analysis of several open-source and industrial LLMs, we observe that models are not reliably better at discriminating among previously-generated alternatives tha"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.04298","kind":"arxiv","version":3},"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/2404.04298/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":"2404.04298","created_at":"2026-07-05T09:03:50.920243+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.04298v3","created_at":"2026-07-05T09:03:50.920243+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.04298","created_at":"2026-07-05T09:03:50.920243+00:00"},{"alias_kind":"pith_short_12","alias_value":"YYLIFSUQRJMM","created_at":"2026-07-05T09:03:50.920243+00:00"},{"alias_kind":"pith_short_16","alias_value":"YYLIFSUQRJMM2PUX","created_at":"2026-07-05T09:03:50.920243+00:00"},{"alias_kind":"pith_short_8","alias_value":"YYLIFSUQ","created_at":"2026-07-05T09:03:50.920243+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22873","citing_title":"SingGuard: A Policy-Adaptive Multimodal LLM Guardrail with Dynamic Reasoning","ref_index":216,"is_internal_anchor":false},{"citing_arxiv_id":"2606.22873","citing_title":"SingGuard: A Policy-Adaptive Multimodal LLM Guardrail with Dynamic Reasoning","ref_index":215,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YYLIFSUQRJMM2PUX7F3FOBH56O","json":"https://pith.science/pith/YYLIFSUQRJMM2PUX7F3FOBH56O.json","graph_json":"https://pith.science/api/pith-number/YYLIFSUQRJMM2PUX7F3FOBH56O/graph.json","events_json":"https://pith.science/api/pith-number/YYLIFSUQRJMM2PUX7F3FOBH56O/events.json","paper":"https://pith.science/paper/YYLIFSUQ"},"agent_actions":{"view_html":"https://pith.science/pith/YYLIFSUQRJMM2PUX7F3FOBH56O","download_json":"https://pith.science/pith/YYLIFSUQRJMM2PUX7F3FOBH56O.json","view_paper":"https://pith.science/paper/YYLIFSUQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.04298&json=true","fetch_graph":"https://pith.science/api/pith-number/YYLIFSUQRJMM2PUX7F3FOBH56O/graph.json","fetch_events":"https://pith.science/api/pith-number/YYLIFSUQRJMM2PUX7F3FOBH56O/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YYLIFSUQRJMM2PUX7F3FOBH56O/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YYLIFSUQRJMM2PUX7F3FOBH56O/action/storage_attestation","attest_author":"https://pith.science/pith/YYLIFSUQRJMM2PUX7F3FOBH56O/action/author_attestation","sign_citation":"https://pith.science/pith/YYLIFSUQRJMM2PUX7F3FOBH56O/action/citation_signature","submit_replication":"https://pith.science/pith/YYLIFSUQRJMM2PUX7F3FOBH56O/action/replication_record"}},"created_at":"2026-07-05T09:03:50.920243+00:00","updated_at":"2026-07-05T09:03:50.920243+00:00"}