{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:XLRF53AKS2YAOONAOITGCEMMDY","short_pith_number":"pith:XLRF53AK","schema_version":"1.0","canonical_sha256":"bae25eec0a96b00739a0722661118c1e24feb62c332421e90c55ca045ff13424","source":{"kind":"arxiv","id":"2311.07468","version":3},"attestation_state":"computed","paper":{"title":"An Analysis and Mitigation of the Reversal Curse","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ang Lv, Ji-Rong Wen, Kaiyi Zhang, Quan Tu, Rui Yan, Shufang Xie, Yuhan Chen","submitted_at":"2023-11-13T17:01:12Z","abstract_excerpt":"Recent research observed a noteworthy phenomenon in large language models (LLMs), referred to as the ``reversal curse.'' The reversal curse is that when dealing with two entities, denoted as $a$ and $b$, connected by their relation $R$ and its inverse $R^{-1}$, LLMs excel in handling sequences in the form of ``$aRb$,'' but encounter challenges when processing ``$bR^{-1}a$,'' whether in generation or comprehension. For instance, GPT-4 can accurately respond to the query ``Tom Cruise's mother is?'' with ``Mary Lee Pfeiffer,'' but it struggles to provide a satisfactory answer when asked ``Mary Le"},"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":"2311.07468","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-11-13T17:01:12Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"4c751b573c91b05766cc85956d0d29e5745986c08c9545070bfc4e97f650123c","abstract_canon_sha256":"71e45842c60876e6e00642a22c59978655de867d838d77c51d8de6d90f306488"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:33:26.656064Z","signature_b64":"EKAzkrvWVv+VNUdFrfbB5NYeDadwDe7zjLUQvUbwesxv2oPDyv2JRr4W9TN/8HEOX04/xlFgHlWlGLCwz7Z0Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bae25eec0a96b00739a0722661118c1e24feb62c332421e90c55ca045ff13424","last_reissued_at":"2026-07-05T09:33:26.655553Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:33:26.655553Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"An Analysis and Mitigation of the Reversal Curse","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ang Lv, Ji-Rong Wen, Kaiyi Zhang, Quan Tu, Rui Yan, Shufang Xie, Yuhan Chen","submitted_at":"2023-11-13T17:01:12Z","abstract_excerpt":"Recent research observed a noteworthy phenomenon in large language models (LLMs), referred to as the ``reversal curse.'' The reversal curse is that when dealing with two entities, denoted as $a$ and $b$, connected by their relation $R$ and its inverse $R^{-1}$, LLMs excel in handling sequences in the form of ``$aRb$,'' but encounter challenges when processing ``$bR^{-1}a$,'' whether in generation or comprehension. For instance, GPT-4 can accurately respond to the query ``Tom Cruise's mother is?'' with ``Mary Lee Pfeiffer,'' but it struggles to provide a satisfactory answer when asked ``Mary Le"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.07468","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/2311.07468/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":"2311.07468","created_at":"2026-07-05T09:33:26.655612+00:00"},{"alias_kind":"arxiv_version","alias_value":"2311.07468v3","created_at":"2026-07-05T09:33:26.655612+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.07468","created_at":"2026-07-05T09:33:26.655612+00:00"},{"alias_kind":"pith_short_12","alias_value":"XLRF53AKS2YA","created_at":"2026-07-05T09:33:26.655612+00:00"},{"alias_kind":"pith_short_16","alias_value":"XLRF53AKS2YAOONA","created_at":"2026-07-05T09:33:26.655612+00:00"},{"alias_kind":"pith_short_8","alias_value":"XLRF53AK","created_at":"2026-07-05T09:33:26.655612+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25476","citing_title":"A Red Teaming Framework for Large Language Models: A Case Study on Faithfulness Evaluation","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27361","citing_title":"Autoregressive Boltzmann Generators","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2406.03736","citing_title":"Your Absorbing Discrete Diffusion Secretly Models the Conditional Distributions of Clean Data","ref_index":73,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06885","citing_title":"Don't Retrain, Align: Adapting Autoregressive LMs to Diffusion LMs via Representation Alignment","ref_index":35,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XLRF53AKS2YAOONAOITGCEMMDY","json":"https://pith.science/pith/XLRF53AKS2YAOONAOITGCEMMDY.json","graph_json":"https://pith.science/api/pith-number/XLRF53AKS2YAOONAOITGCEMMDY/graph.json","events_json":"https://pith.science/api/pith-number/XLRF53AKS2YAOONAOITGCEMMDY/events.json","paper":"https://pith.science/paper/XLRF53AK"},"agent_actions":{"view_html":"https://pith.science/pith/XLRF53AKS2YAOONAOITGCEMMDY","download_json":"https://pith.science/pith/XLRF53AKS2YAOONAOITGCEMMDY.json","view_paper":"https://pith.science/paper/XLRF53AK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2311.07468&json=true","fetch_graph":"https://pith.science/api/pith-number/XLRF53AKS2YAOONAOITGCEMMDY/graph.json","fetch_events":"https://pith.science/api/pith-number/XLRF53AKS2YAOONAOITGCEMMDY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XLRF53AKS2YAOONAOITGCEMMDY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XLRF53AKS2YAOONAOITGCEMMDY/action/storage_attestation","attest_author":"https://pith.science/pith/XLRF53AKS2YAOONAOITGCEMMDY/action/author_attestation","sign_citation":"https://pith.science/pith/XLRF53AKS2YAOONAOITGCEMMDY/action/citation_signature","submit_replication":"https://pith.science/pith/XLRF53AKS2YAOONAOITGCEMMDY/action/replication_record"}},"created_at":"2026-07-05T09:33:26.655612+00:00","updated_at":"2026-07-05T09:33:26.655612+00:00"}