{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:YMVAF3HBN4V4DT3EZN2JEUBZCF","short_pith_number":"pith:YMVAF3HB","schema_version":"1.0","canonical_sha256":"c32a02ece16f2bc1cf64cb7492503911737326a21ed3daccfc4a8bb2e9aba273","source":{"kind":"arxiv","id":"2506.00845","version":3},"attestation_state":"computed","paper":{"title":"Generalizable LLM Learning of Graph Synthetic Data with Post-training Alignment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Heng Wang, Shangbin Feng, Xinyun Liu, Yizhuo Zhang, Yulia Tsvetkov, Zhaoxuan Tan","submitted_at":"2025-06-01T05:39:56Z","abstract_excerpt":"Previous research has sought to enhance the graph reasoning capabilities of LLMs by supervised fine-tuning on synthetic graph data. While these led to specialized LLMs better at solving graph algorithm problems, we don't need LLMs for shortest path: we need generalization from synthetic graph data to real-world tasks with implicit graph structures. In this work, we propose to unlock generalizable learning of graph with post-training alignment with synthetic data. We first design solution-based and process-based rewards for synthetic graph problems: instead of rigid memorizing response patterns"},"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":"2506.00845","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-06-01T05:39:56Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"f8d9bf872d0065095437da41dca378b6cf2ae6d661a20474b12f026f00e815af","abstract_canon_sha256":"944afe8920f5e0b0af930f055bc6f0a58d61b8ac8f61e4a3f5d8634f3b2b7466"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:54:47.716559Z","signature_b64":"1vQpnri5DUumYhwdXIZ2NidYVoGxhrt1hT7PwtknO4RnyRC+vyLEbkEMBRKuYRPfXcqD22Tzn0l2DKX1sRO1CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c32a02ece16f2bc1cf64cb7492503911737326a21ed3daccfc4a8bb2e9aba273","last_reissued_at":"2026-07-05T11:54:47.716066Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:54:47.716066Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Generalizable LLM Learning of Graph Synthetic Data with Post-training Alignment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Heng Wang, Shangbin Feng, Xinyun Liu, Yizhuo Zhang, Yulia Tsvetkov, Zhaoxuan Tan","submitted_at":"2025-06-01T05:39:56Z","abstract_excerpt":"Previous research has sought to enhance the graph reasoning capabilities of LLMs by supervised fine-tuning on synthetic graph data. While these led to specialized LLMs better at solving graph algorithm problems, we don't need LLMs for shortest path: we need generalization from synthetic graph data to real-world tasks with implicit graph structures. In this work, we propose to unlock generalizable learning of graph with post-training alignment with synthetic data. We first design solution-based and process-based rewards for synthetic graph problems: instead of rigid memorizing response patterns"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.00845","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/2506.00845/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":"2506.00845","created_at":"2026-07-05T11:54:47.716126+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.00845v3","created_at":"2026-07-05T11:54:47.716126+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.00845","created_at":"2026-07-05T11:54:47.716126+00:00"},{"alias_kind":"pith_short_12","alias_value":"YMVAF3HBN4V4","created_at":"2026-07-05T11:54:47.716126+00:00"},{"alias_kind":"pith_short_16","alias_value":"YMVAF3HBN4V4DT3E","created_at":"2026-07-05T11:54:47.716126+00:00"},{"alias_kind":"pith_short_8","alias_value":"YMVAF3HB","created_at":"2026-07-05T11:54:47.716126+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25363","citing_title":"TheoremGraph: Bridging Formal and Informal Mathematics","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06865","citing_title":"Are Large Language Models Suitable for Graph Computation? Progress and Prospects","ref_index":265,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YMVAF3HBN4V4DT3EZN2JEUBZCF","json":"https://pith.science/pith/YMVAF3HBN4V4DT3EZN2JEUBZCF.json","graph_json":"https://pith.science/api/pith-number/YMVAF3HBN4V4DT3EZN2JEUBZCF/graph.json","events_json":"https://pith.science/api/pith-number/YMVAF3HBN4V4DT3EZN2JEUBZCF/events.json","paper":"https://pith.science/paper/YMVAF3HB"},"agent_actions":{"view_html":"https://pith.science/pith/YMVAF3HBN4V4DT3EZN2JEUBZCF","download_json":"https://pith.science/pith/YMVAF3HBN4V4DT3EZN2JEUBZCF.json","view_paper":"https://pith.science/paper/YMVAF3HB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.00845&json=true","fetch_graph":"https://pith.science/api/pith-number/YMVAF3HBN4V4DT3EZN2JEUBZCF/graph.json","fetch_events":"https://pith.science/api/pith-number/YMVAF3HBN4V4DT3EZN2JEUBZCF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YMVAF3HBN4V4DT3EZN2JEUBZCF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YMVAF3HBN4V4DT3EZN2JEUBZCF/action/storage_attestation","attest_author":"https://pith.science/pith/YMVAF3HBN4V4DT3EZN2JEUBZCF/action/author_attestation","sign_citation":"https://pith.science/pith/YMVAF3HBN4V4DT3EZN2JEUBZCF/action/citation_signature","submit_replication":"https://pith.science/pith/YMVAF3HBN4V4DT3EZN2JEUBZCF/action/replication_record"}},"created_at":"2026-07-05T11:54:47.716126+00:00","updated_at":"2026-07-05T11:54:47.716126+00:00"}