{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:2LPLKVGTVT425IORBIYTBYB3NA","short_pith_number":"pith:2LPLKVGT","schema_version":"1.0","canonical_sha256":"d2deb554d3acf9aea1d10a3130e03b6820eff0653a7bb5c38a1de9f49faab699","source":{"kind":"arxiv","id":"2411.08302","version":2},"attestation_state":"computed","paper":{"title":"RED: Unleashing Token-Level Rewards from Holistic Feedback via Reward Redistribution","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Cheng Yang, Jiahui Li, Jun Zhou, Kun Kuang, Lin Li, Long Chen, Tai-Wei Chang","submitted_at":"2024-11-13T02:45:21Z","abstract_excerpt":"Reinforcement learning from human feedback (RLHF) offers a promising approach to aligning large language models (LLMs) with human preferences. Typically, a reward model is trained or supplied to act as a proxy for humans in evaluating generated responses during the reinforcement training phase. However, current reward models operate as sequence-to-one models, allocating a single, sparse, and delayed reward to an entire output sequence. This approach may overlook the significant contributions of individual tokens toward the desired outcome. To this end, we propose a more fine-grained, token-lev"},"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":"2411.08302","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-11-13T02:45:21Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e5acba440d9e98e9efb668039ba3a62991f0b6c27f3fc6d6cd7c04d8a5873b0e","abstract_canon_sha256":"22ac963621ca784df0bb899803a79787833f4e01b6e4136604344979e95b34c4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:09:11.411725Z","signature_b64":"l1qe52xlbQ9ExqRvy0yk/2RJuN1Uq70QRmsdxQN2Y8QqaizbJiXU836i3jwyu6gWqwlVAkjEeqv3TbPlBFqrCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d2deb554d3acf9aea1d10a3130e03b6820eff0653a7bb5c38a1de9f49faab699","last_reissued_at":"2026-07-05T12:09:11.411236Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:09:11.411236Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RED: Unleashing Token-Level Rewards from Holistic Feedback via Reward Redistribution","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Cheng Yang, Jiahui Li, Jun Zhou, Kun Kuang, Lin Li, Long Chen, Tai-Wei Chang","submitted_at":"2024-11-13T02:45:21Z","abstract_excerpt":"Reinforcement learning from human feedback (RLHF) offers a promising approach to aligning large language models (LLMs) with human preferences. Typically, a reward model is trained or supplied to act as a proxy for humans in evaluating generated responses during the reinforcement training phase. However, current reward models operate as sequence-to-one models, allocating a single, sparse, and delayed reward to an entire output sequence. This approach may overlook the significant contributions of individual tokens toward the desired outcome. To this end, we propose a more fine-grained, token-lev"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.08302","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/2411.08302/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":"2411.08302","created_at":"2026-07-05T12:09:11.411293+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.08302v2","created_at":"2026-07-05T12:09:11.411293+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.08302","created_at":"2026-07-05T12:09:11.411293+00:00"},{"alias_kind":"pith_short_12","alias_value":"2LPLKVGTVT42","created_at":"2026-07-05T12:09:11.411293+00:00"},{"alias_kind":"pith_short_16","alias_value":"2LPLKVGTVT425IOR","created_at":"2026-07-05T12:09:11.411293+00:00"},{"alias_kind":"pith_short_8","alias_value":"2LPLKVGT","created_at":"2026-07-05T12:09:11.411293+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2LPLKVGTVT425IORBIYTBYB3NA","json":"https://pith.science/pith/2LPLKVGTVT425IORBIYTBYB3NA.json","graph_json":"https://pith.science/api/pith-number/2LPLKVGTVT425IORBIYTBYB3NA/graph.json","events_json":"https://pith.science/api/pith-number/2LPLKVGTVT425IORBIYTBYB3NA/events.json","paper":"https://pith.science/paper/2LPLKVGT"},"agent_actions":{"view_html":"https://pith.science/pith/2LPLKVGTVT425IORBIYTBYB3NA","download_json":"https://pith.science/pith/2LPLKVGTVT425IORBIYTBYB3NA.json","view_paper":"https://pith.science/paper/2LPLKVGT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.08302&json=true","fetch_graph":"https://pith.science/api/pith-number/2LPLKVGTVT425IORBIYTBYB3NA/graph.json","fetch_events":"https://pith.science/api/pith-number/2LPLKVGTVT425IORBIYTBYB3NA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2LPLKVGTVT425IORBIYTBYB3NA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2LPLKVGTVT425IORBIYTBYB3NA/action/storage_attestation","attest_author":"https://pith.science/pith/2LPLKVGTVT425IORBIYTBYB3NA/action/author_attestation","sign_citation":"https://pith.science/pith/2LPLKVGTVT425IORBIYTBYB3NA/action/citation_signature","submit_replication":"https://pith.science/pith/2LPLKVGTVT425IORBIYTBYB3NA/action/replication_record"}},"created_at":"2026-07-05T12:09:11.411293+00:00","updated_at":"2026-07-05T12:09:11.411293+00:00"}