{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FTP66QOF4PYVBPQ4OVE4223PF4","short_pith_number":"pith:FTP66QOF","schema_version":"1.0","canonical_sha256":"2cdfef41c5e3f150be1c7549cd6b6f2f077bdad3b63ab0c367ae9c7fee4d9988","source":{"kind":"arxiv","id":"2503.02368","version":4},"attestation_state":"computed","paper":{"title":"Evolutionary Guided Decoding: Iterative Value Refinement for LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Jing Shao, Lijun Li, Ruizhe Chen, Tong Zhu, Wenliang Chen, Yuxian Jiang, Zhaochen Su, Zhenhua Liu","submitted_at":"2025-03-04T07:49:10Z","abstract_excerpt":"While guided decoding, especially value-guided methods, has emerged as a cost-effective alternative for controlling language model outputs without re-training models, its effectiveness is limited by the accuracy of the value function. We identify that this inaccuracy stems from a core distributional gap: existing methods train static value functions on trajectories sampled exclusively from the base policy, which inherently confines their training to a narrow and suboptimal view of the potential output space. We propose Iterative Value Refinement, a evolutionary framework designed to narrow thi"},"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":"2503.02368","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-04T07:49:10Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"269dbe2b9d1210d65ceff48af656224ee7a6cfa2b1d56777479b63fbafa0e4a9","abstract_canon_sha256":"7a97d63d9b7930257ff7e3f7bbf8accfcfe8daf7d4e7a43c7f4f4ff935e2d0dd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:19:37.539288Z","signature_b64":"L4NDiUyQ0XSBCVJ0XWrU3Ha9B7+kBQnfeyTwsDGkhHlObEa+N/R3I5G0WEAqdlUOsKJL5/pOAFFUT2l07tztAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2cdfef41c5e3f150be1c7549cd6b6f2f077bdad3b63ab0c367ae9c7fee4d9988","last_reissued_at":"2026-07-07T02:19:37.538367Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:19:37.538367Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Evolutionary Guided Decoding: Iterative Value Refinement for LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Jing Shao, Lijun Li, Ruizhe Chen, Tong Zhu, Wenliang Chen, Yuxian Jiang, Zhaochen Su, Zhenhua Liu","submitted_at":"2025-03-04T07:49:10Z","abstract_excerpt":"While guided decoding, especially value-guided methods, has emerged as a cost-effective alternative for controlling language model outputs without re-training models, its effectiveness is limited by the accuracy of the value function. We identify that this inaccuracy stems from a core distributional gap: existing methods train static value functions on trajectories sampled exclusively from the base policy, which inherently confines their training to a narrow and suboptimal view of the potential output space. We propose Iterative Value Refinement, a evolutionary framework designed to narrow thi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.02368","kind":"arxiv","version":4},"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/2503.02368/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":"2503.02368","created_at":"2026-07-07T02:19:37.538485+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.02368v4","created_at":"2026-07-07T02:19:37.538485+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.02368","created_at":"2026-07-07T02:19:37.538485+00:00"},{"alias_kind":"pith_short_12","alias_value":"FTP66QOF4PYV","created_at":"2026-07-07T02:19:37.538485+00:00"},{"alias_kind":"pith_short_16","alias_value":"FTP66QOF4PYVBPQ4","created_at":"2026-07-07T02:19:37.538485+00:00"},{"alias_kind":"pith_short_8","alias_value":"FTP66QOF","created_at":"2026-07-07T02:19:37.538485+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/FTP66QOF4PYVBPQ4OVE4223PF4","json":"https://pith.science/pith/FTP66QOF4PYVBPQ4OVE4223PF4.json","graph_json":"https://pith.science/api/pith-number/FTP66QOF4PYVBPQ4OVE4223PF4/graph.json","events_json":"https://pith.science/api/pith-number/FTP66QOF4PYVBPQ4OVE4223PF4/events.json","paper":"https://pith.science/paper/FTP66QOF"},"agent_actions":{"view_html":"https://pith.science/pith/FTP66QOF4PYVBPQ4OVE4223PF4","download_json":"https://pith.science/pith/FTP66QOF4PYVBPQ4OVE4223PF4.json","view_paper":"https://pith.science/paper/FTP66QOF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.02368&json=true","fetch_graph":"https://pith.science/api/pith-number/FTP66QOF4PYVBPQ4OVE4223PF4/graph.json","fetch_events":"https://pith.science/api/pith-number/FTP66QOF4PYVBPQ4OVE4223PF4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FTP66QOF4PYVBPQ4OVE4223PF4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FTP66QOF4PYVBPQ4OVE4223PF4/action/storage_attestation","attest_author":"https://pith.science/pith/FTP66QOF4PYVBPQ4OVE4223PF4/action/author_attestation","sign_citation":"https://pith.science/pith/FTP66QOF4PYVBPQ4OVE4223PF4/action/citation_signature","submit_replication":"https://pith.science/pith/FTP66QOF4PYVBPQ4OVE4223PF4/action/replication_record"}},"created_at":"2026-07-07T02:19:37.538485+00:00","updated_at":"2026-07-07T02:19:37.538485+00:00"}