{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:QTNYVKAXG44QG34ADQKGWUHFLO","short_pith_number":"pith:QTNYVKAX","schema_version":"1.0","canonical_sha256":"84db8aa8173739036f801c146b50e55b8c7612c3d352a6295ebc739569ee4939","source":{"kind":"arxiv","id":"2406.15992","version":2},"attestation_state":"computed","paper":{"title":"Can LLM Graph Reasoning Generalize beyond Pattern Memorization?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Heng Wang, Shangbin Feng, Tianxing He, Xiaochuang Han, Yizhuo Zhang, Yulia Tsvetkov, Zhaoxuan Tan","submitted_at":"2024-06-23T02:59:15Z","abstract_excerpt":"Large language models (LLMs) demonstrate great potential for problems with implicit graphical structures, while recent works seek to enhance the graph reasoning capabilities of LLMs through specialized instruction tuning. The resulting 'graph LLMs' are evaluated with in-distribution settings only, thus it remains underexplored whether LLMs are learning generalizable graph reasoning skills or merely memorizing patterns in the synthetic training data. To this end, we propose the NLGift benchmark, an evaluation suite of LLM graph reasoning generalization: whether LLMs could go beyond semantic, nu"},"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":"2406.15992","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-23T02:59:15Z","cross_cats_sorted":[],"title_canon_sha256":"a797edf9986887ec7afaf56b0c8d533716ad21f271f09096ebd9f06a0bb91b11","abstract_canon_sha256":"65c47ef272679db3e557043ee9d1590b94bf9d04b9d3009fbbe7fd21114bc124"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:19:10.669067Z","signature_b64":"d4R96F1YhRSZHbXoBAgdGUGsdxqRUJz9qWHgxcUUFmNQt2x7F3oksM1CuJsg44zL+9V9OYPJ2HPsl7vTMHkCDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"84db8aa8173739036f801c146b50e55b8c7612c3d352a6295ebc739569ee4939","last_reissued_at":"2026-07-05T09:19:10.668504Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:19:10.668504Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Can LLM Graph Reasoning Generalize beyond Pattern Memorization?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Heng Wang, Shangbin Feng, Tianxing He, Xiaochuang Han, Yizhuo Zhang, Yulia Tsvetkov, Zhaoxuan Tan","submitted_at":"2024-06-23T02:59:15Z","abstract_excerpt":"Large language models (LLMs) demonstrate great potential for problems with implicit graphical structures, while recent works seek to enhance the graph reasoning capabilities of LLMs through specialized instruction tuning. The resulting 'graph LLMs' are evaluated with in-distribution settings only, thus it remains underexplored whether LLMs are learning generalizable graph reasoning skills or merely memorizing patterns in the synthetic training data. To this end, we propose the NLGift benchmark, an evaluation suite of LLM graph reasoning generalization: whether LLMs could go beyond semantic, nu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.15992","kind":"arxiv","version":2},"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/2406.15992/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":"2406.15992","created_at":"2026-07-05T09:19:10.668575+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.15992v2","created_at":"2026-07-05T09:19:10.668575+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.15992","created_at":"2026-07-05T09:19:10.668575+00:00"},{"alias_kind":"pith_short_12","alias_value":"QTNYVKAXG44Q","created_at":"2026-07-05T09:19:10.668575+00:00"},{"alias_kind":"pith_short_16","alias_value":"QTNYVKAXG44QG34A","created_at":"2026-07-05T09:19:10.668575+00:00"},{"alias_kind":"pith_short_8","alias_value":"QTNYVKAX","created_at":"2026-07-05T09:19:10.668575+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.31404","citing_title":"The Sword, Shield, and Achilles' Heel: Characterizing the Linguistic Inductive Bias of Large Language Models for Spatial Reasoning in Navigation Planning","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2512.17321","citing_title":"Neuro-Symbolic Control with Large Language Models for Language-Guided Spatial Tasks","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15306","citing_title":"Generalization in LLM Problem Solving: The Case of the Shortest Path","ref_index":54,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QTNYVKAXG44QG34ADQKGWUHFLO","json":"https://pith.science/pith/QTNYVKAXG44QG34ADQKGWUHFLO.json","graph_json":"https://pith.science/api/pith-number/QTNYVKAXG44QG34ADQKGWUHFLO/graph.json","events_json":"https://pith.science/api/pith-number/QTNYVKAXG44QG34ADQKGWUHFLO/events.json","paper":"https://pith.science/paper/QTNYVKAX"},"agent_actions":{"view_html":"https://pith.science/pith/QTNYVKAXG44QG34ADQKGWUHFLO","download_json":"https://pith.science/pith/QTNYVKAXG44QG34ADQKGWUHFLO.json","view_paper":"https://pith.science/paper/QTNYVKAX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.15992&json=true","fetch_graph":"https://pith.science/api/pith-number/QTNYVKAXG44QG34ADQKGWUHFLO/graph.json","fetch_events":"https://pith.science/api/pith-number/QTNYVKAXG44QG34ADQKGWUHFLO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QTNYVKAXG44QG34ADQKGWUHFLO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QTNYVKAXG44QG34ADQKGWUHFLO/action/storage_attestation","attest_author":"https://pith.science/pith/QTNYVKAXG44QG34ADQKGWUHFLO/action/author_attestation","sign_citation":"https://pith.science/pith/QTNYVKAXG44QG34ADQKGWUHFLO/action/citation_signature","submit_replication":"https://pith.science/pith/QTNYVKAXG44QG34ADQKGWUHFLO/action/replication_record"}},"created_at":"2026-07-05T09:19:10.668575+00:00","updated_at":"2026-07-05T09:19:10.668575+00:00"}