{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KASNF2BD2GEL4JKWKQXMPEP4WK","short_pith_number":"pith:KASNF2BD","schema_version":"1.0","canonical_sha256":"5024d2e823d188be2556542ec791fcb2a076296eb1cf295c115153babe7fbded","source":{"kind":"arxiv","id":"2506.15651","version":1},"attestation_state":"computed","paper":{"title":"AutoRule: Reasoning Chain-of-thought Extracted Rule-based Rewards Improve Preference Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Chenyan Xiong, Tevin Wang","submitted_at":"2025-06-18T17:29:19Z","abstract_excerpt":"Rule-based rewards offer a promising strategy for improving reinforcement learning from human feedback (RLHF), but current approaches often rely on manual rule engineering. We present AutoRule, a fully automated method for extracting rules from preference feedback and formulating them into rule-based rewards. AutoRule extraction operates in three stages: it leverages a reasoning model to interpret user preferences, identifies candidate rules from the reasoning chain of these interpretations, and synthesizes them into a unified rule set. Leveraging the finalized rule set, we employ language-mod"},"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":"2506.15651","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-06-18T17:29:19Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"900998390ae6cae888b3c4c5af72fd121172d2d1e3c79a68a9afc6b5b61628fd","abstract_canon_sha256":"62deaa634e36128c27600bda6451ada8380e828d54309c5b5f077b40fbe63afb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:23:47.796441Z","signature_b64":"YTam8Xp6VWK7n2lWTd+oceRi123bn4d4KbBZ9x1w7BSueJzN26BmwFGI4vygz6E6DlklDUDs6i76klBy5+VHBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5024d2e823d188be2556542ec791fcb2a076296eb1cf295c115153babe7fbded","last_reissued_at":"2026-07-05T11:23:47.795893Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:23:47.795893Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AutoRule: Reasoning Chain-of-thought Extracted Rule-based Rewards Improve Preference Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Chenyan Xiong, Tevin Wang","submitted_at":"2025-06-18T17:29:19Z","abstract_excerpt":"Rule-based rewards offer a promising strategy for improving reinforcement learning from human feedback (RLHF), but current approaches often rely on manual rule engineering. We present AutoRule, a fully automated method for extracting rules from preference feedback and formulating them into rule-based rewards. AutoRule extraction operates in three stages: it leverages a reasoning model to interpret user preferences, identifies candidate rules from the reasoning chain of these interpretations, and synthesizes them into a unified rule set. Leveraging the finalized rule set, we employ language-mod"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.15651","kind":"arxiv","version":1},"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/2506.15651/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":"2506.15651","created_at":"2026-07-05T11:23:47.795956+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.15651v1","created_at":"2026-07-05T11:23:47.795956+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.15651","created_at":"2026-07-05T11:23:47.795956+00:00"},{"alias_kind":"pith_short_12","alias_value":"KASNF2BD2GEL","created_at":"2026-07-05T11:23:47.795956+00:00"},{"alias_kind":"pith_short_16","alias_value":"KASNF2BD2GEL4JKW","created_at":"2026-07-05T11:23:47.795956+00:00"},{"alias_kind":"pith_short_8","alias_value":"KASNF2BD","created_at":"2026-07-05T11:23:47.795956+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21943","citing_title":"Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning","ref_index":219,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30568","citing_title":"Generating and Refining Dynamic Evaluation Rubrics for LLM-as-a-Judge","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17602","citing_title":"AutoRubric-T2I: Robust Rule-Based Reward Model for Text-to-Image Alignment","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17602","citing_title":"AutoRubric-T2I: Robust Rule-Based Reward Model for Text-to-Image Alignment","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2507.21046","citing_title":"A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence","ref_index":262,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19341","citing_title":"Evaluation-driven Scaling for Scientific Discovery","ref_index":151,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18131","citing_title":"Training LLM Agents for Spontaneous, Reward-Free Self-Evolution via World Knowledge Exploration","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KASNF2BD2GEL4JKWKQXMPEP4WK","json":"https://pith.science/pith/KASNF2BD2GEL4JKWKQXMPEP4WK.json","graph_json":"https://pith.science/api/pith-number/KASNF2BD2GEL4JKWKQXMPEP4WK/graph.json","events_json":"https://pith.science/api/pith-number/KASNF2BD2GEL4JKWKQXMPEP4WK/events.json","paper":"https://pith.science/paper/KASNF2BD"},"agent_actions":{"view_html":"https://pith.science/pith/KASNF2BD2GEL4JKWKQXMPEP4WK","download_json":"https://pith.science/pith/KASNF2BD2GEL4JKWKQXMPEP4WK.json","view_paper":"https://pith.science/paper/KASNF2BD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.15651&json=true","fetch_graph":"https://pith.science/api/pith-number/KASNF2BD2GEL4JKWKQXMPEP4WK/graph.json","fetch_events":"https://pith.science/api/pith-number/KASNF2BD2GEL4JKWKQXMPEP4WK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KASNF2BD2GEL4JKWKQXMPEP4WK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KASNF2BD2GEL4JKWKQXMPEP4WK/action/storage_attestation","attest_author":"https://pith.science/pith/KASNF2BD2GEL4JKWKQXMPEP4WK/action/author_attestation","sign_citation":"https://pith.science/pith/KASNF2BD2GEL4JKWKQXMPEP4WK/action/citation_signature","submit_replication":"https://pith.science/pith/KASNF2BD2GEL4JKWKQXMPEP4WK/action/replication_record"}},"created_at":"2026-07-05T11:23:47.795956+00:00","updated_at":"2026-07-05T11:23:47.795956+00:00"}