{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:BXRAWZ5LEOZBYTY3SFEJOPHQHL","short_pith_number":"pith:BXRAWZ5L","schema_version":"1.0","canonical_sha256":"0de20b67ab23b21c4f1b9148973cf03ad6c15116b5003b6726554ddf25114b86","source":{"kind":"arxiv","id":"2510.17385","version":5},"attestation_state":"computed","paper":{"title":"Strengthening LLMs for Tabular Prediction with Structural Priors","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Guocong Li, Jintai Chen, Pengxiang Cai, Wanchen Lian, Zihao Gao","submitted_at":"2025-10-20T10:22:01Z","abstract_excerpt":"Tabular prediction has long been dominated by gradient-boosted decision trees and specialized deep tabular models, while large language models (LLMs) remain difficult to make competitive despite their cross-task adaptability and transparent reasoning traces. We address this gap by incorporating tabular structural priors into LLM post-training. Specifically, we propose Permutation Relative Policy Optimization (PRPO), which operationalizes column-permutation invariance through label-preserving column permutations and two-level advantage estimation. This design converts sparse outcome rewards int"},"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":"2510.17385","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-10-20T10:22:01Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"3a6f6fb6b3921fc1896f89ac8edfd3b8e141a1c6178b5ef2f7f6cbcd7e93a442","abstract_canon_sha256":"4d6a4ed19b1de48081252e6151efa58d55d2c835bb0e3b593267760cf34b8951"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T02:13:17.755291Z","signature_b64":"QvAMbyB4ymrFUGjPTZ50IcVrs9l6x4BMG/b2UWyt4D3KF5OCbNd04+kCVs4+rzMIEFrx/BtZRdtFBTDNxVuxDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0de20b67ab23b21c4f1b9148973cf03ad6c15116b5003b6726554ddf25114b86","last_reissued_at":"2026-06-23T02:13:17.754806Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T02:13:17.754806Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Strengthening LLMs for Tabular Prediction with Structural Priors","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Guocong Li, Jintai Chen, Pengxiang Cai, Wanchen Lian, Zihao Gao","submitted_at":"2025-10-20T10:22:01Z","abstract_excerpt":"Tabular prediction has long been dominated by gradient-boosted decision trees and specialized deep tabular models, while large language models (LLMs) remain difficult to make competitive despite their cross-task adaptability and transparent reasoning traces. We address this gap by incorporating tabular structural priors into LLM post-training. Specifically, we propose Permutation Relative Policy Optimization (PRPO), which operationalizes column-permutation invariance through label-preserving column permutations and two-level advantage estimation. This design converts sparse outcome rewards int"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.17385","kind":"arxiv","version":5},"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/2510.17385/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":"2510.17385","created_at":"2026-06-23T02:13:17.754865+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.17385v5","created_at":"2026-06-23T02:13:17.754865+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.17385","created_at":"2026-06-23T02:13:17.754865+00:00"},{"alias_kind":"pith_short_12","alias_value":"BXRAWZ5LEOZB","created_at":"2026-06-23T02:13:17.754865+00:00"},{"alias_kind":"pith_short_16","alias_value":"BXRAWZ5LEOZBYTY3","created_at":"2026-06-23T02:13:17.754865+00:00"},{"alias_kind":"pith_short_8","alias_value":"BXRAWZ5L","created_at":"2026-06-23T02:13:17.754865+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2604.13392","citing_title":"ReSS: Learning Reasoning Models for Tabular Data Prediction via Symbolic Scaffold","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2604.13392","citing_title":"ReSS: Learning Reasoning Models for Tabular Data Prediction via Symbolic Scaffold","ref_index":10,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BXRAWZ5LEOZBYTY3SFEJOPHQHL","json":"https://pith.science/pith/BXRAWZ5LEOZBYTY3SFEJOPHQHL.json","graph_json":"https://pith.science/api/pith-number/BXRAWZ5LEOZBYTY3SFEJOPHQHL/graph.json","events_json":"https://pith.science/api/pith-number/BXRAWZ5LEOZBYTY3SFEJOPHQHL/events.json","paper":"https://pith.science/paper/BXRAWZ5L"},"agent_actions":{"view_html":"https://pith.science/pith/BXRAWZ5LEOZBYTY3SFEJOPHQHL","download_json":"https://pith.science/pith/BXRAWZ5LEOZBYTY3SFEJOPHQHL.json","view_paper":"https://pith.science/paper/BXRAWZ5L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.17385&json=true","fetch_graph":"https://pith.science/api/pith-number/BXRAWZ5LEOZBYTY3SFEJOPHQHL/graph.json","fetch_events":"https://pith.science/api/pith-number/BXRAWZ5LEOZBYTY3SFEJOPHQHL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BXRAWZ5LEOZBYTY3SFEJOPHQHL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BXRAWZ5LEOZBYTY3SFEJOPHQHL/action/storage_attestation","attest_author":"https://pith.science/pith/BXRAWZ5LEOZBYTY3SFEJOPHQHL/action/author_attestation","sign_citation":"https://pith.science/pith/BXRAWZ5LEOZBYTY3SFEJOPHQHL/action/citation_signature","submit_replication":"https://pith.science/pith/BXRAWZ5LEOZBYTY3SFEJOPHQHL/action/replication_record"}},"created_at":"2026-06-23T02:13:17.754865+00:00","updated_at":"2026-06-23T02:13:17.754865+00:00"}