{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:T2XAG2AXEJKUF7C67KZJTFGSE3","short_pith_number":"pith:T2XAG2AX","schema_version":"1.0","canonical_sha256":"9eae036817225542fc5efab29994d226cb0231c6715a207cea1b23ff3ccbbdf2","source":{"kind":"arxiv","id":"2409.06903","version":1},"attestation_state":"computed","paper":{"title":"Semi-Supervised Reward Modeling via Iterative Self-Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alexandros Papangelis, Han Zhao, Haoxiang Wang, Yifei He, Ziyan Jiang","submitted_at":"2024-09-10T22:57:58Z","abstract_excerpt":"Reward models (RM) capture the values and preferences of humans and play a central role in Reinforcement Learning with Human Feedback (RLHF) to align pretrained large language models (LLMs). Traditionally, training these models relies on extensive human-annotated preference data, which poses significant challenges in terms of scalability and cost. To overcome these limitations, we propose Semi-Supervised Reward Modeling (SSRM), an approach that enhances RM training using unlabeled data. Given an unlabeled dataset, SSRM involves three key iterative steps: pseudo-labeling unlabeled examples, sel"},"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":"2409.06903","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-09-10T22:57:58Z","cross_cats_sorted":[],"title_canon_sha256":"ba0bb67bb9008c5630dce5aeababb8518743b69307f2e5eb00099e85fce02b54","abstract_canon_sha256":"d9af61e3144402905f2e9dfa4b849e8f805df34ba3ebf75883b147fc2c38246e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:05:42.899825Z","signature_b64":"BA8ZR798LeAHpPIG1q6pG28G4TMqQ6kYeWIeeTfoHnRztaxBZaSkpeZ4Zo8NpFDkjQgItxg0XIZ01z3zkWTaCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9eae036817225542fc5efab29994d226cb0231c6715a207cea1b23ff3ccbbdf2","last_reissued_at":"2026-07-05T09:05:42.899485Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:05:42.899485Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Semi-Supervised Reward Modeling via Iterative Self-Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alexandros Papangelis, Han Zhao, Haoxiang Wang, Yifei He, Ziyan Jiang","submitted_at":"2024-09-10T22:57:58Z","abstract_excerpt":"Reward models (RM) capture the values and preferences of humans and play a central role in Reinforcement Learning with Human Feedback (RLHF) to align pretrained large language models (LLMs). Traditionally, training these models relies on extensive human-annotated preference data, which poses significant challenges in terms of scalability and cost. To overcome these limitations, we propose Semi-Supervised Reward Modeling (SSRM), an approach that enhances RM training using unlabeled data. Given an unlabeled dataset, SSRM involves three key iterative steps: pseudo-labeling unlabeled examples, sel"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.06903","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/2409.06903/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":"2409.06903","created_at":"2026-07-05T09:05:42.899542+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.06903v1","created_at":"2026-07-05T09:05:42.899542+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.06903","created_at":"2026-07-05T09:05:42.899542+00:00"},{"alias_kind":"pith_short_12","alias_value":"T2XAG2AXEJKU","created_at":"2026-07-05T09:05:42.899542+00:00"},{"alias_kind":"pith_short_16","alias_value":"T2XAG2AXEJKUF7C6","created_at":"2026-07-05T09:05:42.899542+00:00"},{"alias_kind":"pith_short_8","alias_value":"T2XAG2AX","created_at":"2026-07-05T09:05:42.899542+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2412.15118","citing_title":"Reasoning Through Execution: Unifying Process and Outcome Rewards for Code Generation","ref_index":28,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/T2XAG2AXEJKUF7C67KZJTFGSE3","json":"https://pith.science/pith/T2XAG2AXEJKUF7C67KZJTFGSE3.json","graph_json":"https://pith.science/api/pith-number/T2XAG2AXEJKUF7C67KZJTFGSE3/graph.json","events_json":"https://pith.science/api/pith-number/T2XAG2AXEJKUF7C67KZJTFGSE3/events.json","paper":"https://pith.science/paper/T2XAG2AX"},"agent_actions":{"view_html":"https://pith.science/pith/T2XAG2AXEJKUF7C67KZJTFGSE3","download_json":"https://pith.science/pith/T2XAG2AXEJKUF7C67KZJTFGSE3.json","view_paper":"https://pith.science/paper/T2XAG2AX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.06903&json=true","fetch_graph":"https://pith.science/api/pith-number/T2XAG2AXEJKUF7C67KZJTFGSE3/graph.json","fetch_events":"https://pith.science/api/pith-number/T2XAG2AXEJKUF7C67KZJTFGSE3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/T2XAG2AXEJKUF7C67KZJTFGSE3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/T2XAG2AXEJKUF7C67KZJTFGSE3/action/storage_attestation","attest_author":"https://pith.science/pith/T2XAG2AXEJKUF7C67KZJTFGSE3/action/author_attestation","sign_citation":"https://pith.science/pith/T2XAG2AXEJKUF7C67KZJTFGSE3/action/citation_signature","submit_replication":"https://pith.science/pith/T2XAG2AXEJKUF7C67KZJTFGSE3/action/replication_record"}},"created_at":"2026-07-05T09:05:42.899542+00:00","updated_at":"2026-07-05T09:05:42.899542+00:00"}