{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:PKRDOS4RULFALVX2JHDQLXLZEX","short_pith_number":"pith:PKRDOS4R","schema_version":"1.0","canonical_sha256":"7aa2374b91a2ca05d6fa49c705dd7925f775feb0a493c90e6e524ae78dbf21af","source":{"kind":"arxiv","id":"2305.14279","version":4},"attestation_state":"computed","paper":{"title":"Two Failures of Self-Consistency in the Multi-Step Reasoning of LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alicia Parrish, Angelica Chen, Chen Zhao, Jason Phang, Kyunghyun Cho, Samuel R. Bowman, Vishakh Padmakumar","submitted_at":"2023-05-23T17:25:59Z","abstract_excerpt":"Large language models (LLMs) have achieved widespread success on a variety of in-context few-shot tasks, but this success is typically evaluated via correctness rather than consistency. We argue that self-consistency is an important criteria for valid multi-step reasoning in tasks where the solution is composed of the answers to multiple sub-steps. We propose two types of self-consistency that are particularly important for multi-step reasoning -- hypothetical consistency (a model's ability to predict what its output would be in a hypothetical other context) and compositional consistency (cons"},"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":"2305.14279","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-05-23T17:25:59Z","cross_cats_sorted":[],"title_canon_sha256":"d5a11805e9f24dc442ccbdc8d84c090ecb608a7349cf15f1c6b501e19a9d52be","abstract_canon_sha256":"19b5c3a9b3598e49af6d1677a849745b2327b32d1356fbf1d445f19f7591b8c5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:42:39.818617Z","signature_b64":"kVLCARKVue+yZBLbrEXUXqJ2g45YgaxMboF5hWDbPRKZ4DH9sLmIq+wvkbvv0t/MhGP09rtN0G6RI6CoTQr4Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7aa2374b91a2ca05d6fa49c705dd7925f775feb0a493c90e6e524ae78dbf21af","last_reissued_at":"2026-07-05T07:42:39.818210Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:42:39.818210Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Two Failures of Self-Consistency in the Multi-Step Reasoning of LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alicia Parrish, Angelica Chen, Chen Zhao, Jason Phang, Kyunghyun Cho, Samuel R. Bowman, Vishakh Padmakumar","submitted_at":"2023-05-23T17:25:59Z","abstract_excerpt":"Large language models (LLMs) have achieved widespread success on a variety of in-context few-shot tasks, but this success is typically evaluated via correctness rather than consistency. We argue that self-consistency is an important criteria for valid multi-step reasoning in tasks where the solution is composed of the answers to multiple sub-steps. We propose two types of self-consistency that are particularly important for multi-step reasoning -- hypothetical consistency (a model's ability to predict what its output would be in a hypothetical other context) and compositional consistency (cons"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.14279","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/2305.14279/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":"2305.14279","created_at":"2026-07-05T07:42:39.818263+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.14279v4","created_at":"2026-07-05T07:42:39.818263+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.14279","created_at":"2026-07-05T07:42:39.818263+00:00"},{"alias_kind":"pith_short_12","alias_value":"PKRDOS4RULFA","created_at":"2026-07-05T07:42:39.818263+00:00"},{"alias_kind":"pith_short_16","alias_value":"PKRDOS4RULFALVX2","created_at":"2026-07-05T07:42:39.818263+00:00"},{"alias_kind":"pith_short_8","alias_value":"PKRDOS4R","created_at":"2026-07-05T07:42:39.818263+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25489","citing_title":"When LLM Rationales Become User-Facing: Effects on Trust Perception, Decision-Making, and Gaze Behaviors","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23189","citing_title":"Empirical Bayes Conformal Prediction for Vision and Language Models","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2308.11432","citing_title":"A Survey on Large Language Model based Autonomous Agents","ref_index":162,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PKRDOS4RULFALVX2JHDQLXLZEX","json":"https://pith.science/pith/PKRDOS4RULFALVX2JHDQLXLZEX.json","graph_json":"https://pith.science/api/pith-number/PKRDOS4RULFALVX2JHDQLXLZEX/graph.json","events_json":"https://pith.science/api/pith-number/PKRDOS4RULFALVX2JHDQLXLZEX/events.json","paper":"https://pith.science/paper/PKRDOS4R"},"agent_actions":{"view_html":"https://pith.science/pith/PKRDOS4RULFALVX2JHDQLXLZEX","download_json":"https://pith.science/pith/PKRDOS4RULFALVX2JHDQLXLZEX.json","view_paper":"https://pith.science/paper/PKRDOS4R","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.14279&json=true","fetch_graph":"https://pith.science/api/pith-number/PKRDOS4RULFALVX2JHDQLXLZEX/graph.json","fetch_events":"https://pith.science/api/pith-number/PKRDOS4RULFALVX2JHDQLXLZEX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PKRDOS4RULFALVX2JHDQLXLZEX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PKRDOS4RULFALVX2JHDQLXLZEX/action/storage_attestation","attest_author":"https://pith.science/pith/PKRDOS4RULFALVX2JHDQLXLZEX/action/author_attestation","sign_citation":"https://pith.science/pith/PKRDOS4RULFALVX2JHDQLXLZEX/action/citation_signature","submit_replication":"https://pith.science/pith/PKRDOS4RULFALVX2JHDQLXLZEX/action/replication_record"}},"created_at":"2026-07-05T07:42:39.818263+00:00","updated_at":"2026-07-05T07:42:39.818263+00:00"}