{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:ZVSRX5LNNDEDH7C26I4GOPMJIP","short_pith_number":"pith:ZVSRX5LN","schema_version":"1.0","canonical_sha256":"cd651bf56d68c833fc5af238673d8943ec338dd08aeafaebadf771c1171b7b11","source":{"kind":"arxiv","id":"2405.20495","version":1},"attestation_state":"computed","paper":{"title":"Transfer Q Star: Principled Decoding for LLM Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Amrit Singh Bedi, Dinesh Manocha, Furong Huang, Mengdi Wang, Ming Yin, Soumya Suvra Ghosal, Souradip Chakraborty","submitted_at":"2024-05-30T21:36:12Z","abstract_excerpt":"Aligning foundation models is essential for their safe and trustworthy deployment. However, traditional fine-tuning methods are computationally intensive and require updating billions of model parameters. A promising alternative, alignment via decoding, adjusts the response distribution directly without model updates to maximize a target reward $r$, thus providing a lightweight and adaptable framework for alignment. However, principled decoding methods rely on oracle access to an optimal Q-function ($Q^*$), which is often unavailable in practice. Hence, prior SoTA methods either approximate th"},"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":"2405.20495","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-30T21:36:12Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"74dbb6ce66e8c66aaa6bc7589a1e26a474286c545e41bda18c7c9187211c3f26","abstract_canon_sha256":"2b1aae88a6f23cc8ae3cd374b868019e65645e680d7aa7d3ebd78e7e65b6127c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:25:38.156804Z","signature_b64":"KndiA+aELqY6exBBrD7jvxe3BH4dmvzGWgg+9uYcyeOGhr63Jca1E7b3OlLWu33d+KCELvYo5SA99yY7XSk3Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cd651bf56d68c833fc5af238673d8943ec338dd08aeafaebadf771c1171b7b11","last_reissued_at":"2026-07-05T08:25:38.156346Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:25:38.156346Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Transfer Q Star: Principled Decoding for LLM Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Amrit Singh Bedi, Dinesh Manocha, Furong Huang, Mengdi Wang, Ming Yin, Soumya Suvra Ghosal, Souradip Chakraborty","submitted_at":"2024-05-30T21:36:12Z","abstract_excerpt":"Aligning foundation models is essential for their safe and trustworthy deployment. However, traditional fine-tuning methods are computationally intensive and require updating billions of model parameters. A promising alternative, alignment via decoding, adjusts the response distribution directly without model updates to maximize a target reward $r$, thus providing a lightweight and adaptable framework for alignment. However, principled decoding methods rely on oracle access to an optimal Q-function ($Q^*$), which is often unavailable in practice. Hence, prior SoTA methods either approximate th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.20495","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/2405.20495/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":"2405.20495","created_at":"2026-07-05T08:25:38.156402+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.20495v1","created_at":"2026-07-05T08:25:38.156402+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.20495","created_at":"2026-07-05T08:25:38.156402+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZVSRX5LNNDED","created_at":"2026-07-05T08:25:38.156402+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZVSRX5LNNDEDH7C2","created_at":"2026-07-05T08:25:38.156402+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZVSRX5LN","created_at":"2026-07-05T08:25:38.156402+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2408.08812","citing_title":"TRAM: Test-Time Risk Adaptation with Mixture of Agents","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2507.06419","citing_title":"Teach a Reward Model to Correct Itself: Reward Guided Adversarial Failure Discovery for Robust Reward Modeling","ref_index":4,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZVSRX5LNNDEDH7C26I4GOPMJIP","json":"https://pith.science/pith/ZVSRX5LNNDEDH7C26I4GOPMJIP.json","graph_json":"https://pith.science/api/pith-number/ZVSRX5LNNDEDH7C26I4GOPMJIP/graph.json","events_json":"https://pith.science/api/pith-number/ZVSRX5LNNDEDH7C26I4GOPMJIP/events.json","paper":"https://pith.science/paper/ZVSRX5LN"},"agent_actions":{"view_html":"https://pith.science/pith/ZVSRX5LNNDEDH7C26I4GOPMJIP","download_json":"https://pith.science/pith/ZVSRX5LNNDEDH7C26I4GOPMJIP.json","view_paper":"https://pith.science/paper/ZVSRX5LN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.20495&json=true","fetch_graph":"https://pith.science/api/pith-number/ZVSRX5LNNDEDH7C26I4GOPMJIP/graph.json","fetch_events":"https://pith.science/api/pith-number/ZVSRX5LNNDEDH7C26I4GOPMJIP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZVSRX5LNNDEDH7C26I4GOPMJIP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZVSRX5LNNDEDH7C26I4GOPMJIP/action/storage_attestation","attest_author":"https://pith.science/pith/ZVSRX5LNNDEDH7C26I4GOPMJIP/action/author_attestation","sign_citation":"https://pith.science/pith/ZVSRX5LNNDEDH7C26I4GOPMJIP/action/citation_signature","submit_replication":"https://pith.science/pith/ZVSRX5LNNDEDH7C26I4GOPMJIP/action/replication_record"}},"created_at":"2026-07-05T08:25:38.156402+00:00","updated_at":"2026-07-05T08:25:38.156402+00:00"}