{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4JNQRESDBMV42FCWIEQ7YW6XB3","short_pith_number":"pith:4JNQRESD","schema_version":"1.0","canonical_sha256":"e25b0892430b2bcd14564121fc5bd70ee247d4ad6d2ab564dacbee89d37c445b","source":{"kind":"arxiv","id":"2410.09528","version":2},"attestation_state":"computed","paper":{"title":"Boosting Deductive Reasoning with Step Signals In RLHF","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dong Yan, Jialian Li, Jian Xie, Wei Shen, Yipin Zhang, Yuzi Yan","submitted_at":"2024-10-12T13:19:11Z","abstract_excerpt":"Logical reasoning is a crucial task for Large Language Models (LLMs), enabling them to tackle complex problems. Among reasoning tasks, multi-step reasoning poses a particular challenge. Grounded in the theory of formal logic, we have developed an automated method, Multi-step Deduction (MuseD), for deductive reasoning data. MuseD has allowed us to create training and testing datasets for multi-step reasoning. Our generation method enables control over the complexity of the generated instructions, facilitating training and evaluation of models across different difficulty levels. Through RLHF tra"},"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":"2410.09528","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-10-12T13:19:11Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"226d43d5bdc8469474caed7bbd82523dc5183a657d4ace917ebff6ba64fdf485","abstract_canon_sha256":"b9ddbae7db581a1f8cda5119a81a4557f62659d9f61b46867ae7cb9d25dd19e3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:25:02.150360Z","signature_b64":"w3QbWvWH5vRT33Lp8I+l+IeVlrwyN4N676/xjikkKQpUtDG0lXvGiHcR0292giGpvwrs8rC0wuauDd3EuddEBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e25b0892430b2bcd14564121fc5bd70ee247d4ad6d2ab564dacbee89d37c445b","last_reissued_at":"2026-07-05T09:25:02.149921Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:25:02.149921Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Boosting Deductive Reasoning with Step Signals In RLHF","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dong Yan, Jialian Li, Jian Xie, Wei Shen, Yipin Zhang, Yuzi Yan","submitted_at":"2024-10-12T13:19:11Z","abstract_excerpt":"Logical reasoning is a crucial task for Large Language Models (LLMs), enabling them to tackle complex problems. Among reasoning tasks, multi-step reasoning poses a particular challenge. Grounded in the theory of formal logic, we have developed an automated method, Multi-step Deduction (MuseD), for deductive reasoning data. MuseD has allowed us to create training and testing datasets for multi-step reasoning. Our generation method enables control over the complexity of the generated instructions, facilitating training and evaluation of models across different difficulty levels. Through RLHF tra"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.09528","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/2410.09528/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":"2410.09528","created_at":"2026-07-05T09:25:02.149979+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.09528v2","created_at":"2026-07-05T09:25:02.149979+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.09528","created_at":"2026-07-05T09:25:02.149979+00:00"},{"alias_kind":"pith_short_12","alias_value":"4JNQRESDBMV4","created_at":"2026-07-05T09:25:02.149979+00:00"},{"alias_kind":"pith_short_16","alias_value":"4JNQRESDBMV42FCW","created_at":"2026-07-05T09:25:02.149979+00:00"},{"alias_kind":"pith_short_8","alias_value":"4JNQRESD","created_at":"2026-07-05T09:25:02.149979+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.19966","citing_title":"Learning to Select In-Context Demonstration Preferred by Large Language Model","ref_index":20,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4JNQRESDBMV42FCWIEQ7YW6XB3","json":"https://pith.science/pith/4JNQRESDBMV42FCWIEQ7YW6XB3.json","graph_json":"https://pith.science/api/pith-number/4JNQRESDBMV42FCWIEQ7YW6XB3/graph.json","events_json":"https://pith.science/api/pith-number/4JNQRESDBMV42FCWIEQ7YW6XB3/events.json","paper":"https://pith.science/paper/4JNQRESD"},"agent_actions":{"view_html":"https://pith.science/pith/4JNQRESDBMV42FCWIEQ7YW6XB3","download_json":"https://pith.science/pith/4JNQRESDBMV42FCWIEQ7YW6XB3.json","view_paper":"https://pith.science/paper/4JNQRESD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.09528&json=true","fetch_graph":"https://pith.science/api/pith-number/4JNQRESDBMV42FCWIEQ7YW6XB3/graph.json","fetch_events":"https://pith.science/api/pith-number/4JNQRESDBMV42FCWIEQ7YW6XB3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4JNQRESDBMV42FCWIEQ7YW6XB3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4JNQRESDBMV42FCWIEQ7YW6XB3/action/storage_attestation","attest_author":"https://pith.science/pith/4JNQRESDBMV42FCWIEQ7YW6XB3/action/author_attestation","sign_citation":"https://pith.science/pith/4JNQRESDBMV42FCWIEQ7YW6XB3/action/citation_signature","submit_replication":"https://pith.science/pith/4JNQRESDBMV42FCWIEQ7YW6XB3/action/replication_record"}},"created_at":"2026-07-05T09:25:02.149979+00:00","updated_at":"2026-07-05T09:25:02.149979+00:00"}