{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:AZU3KX4SO3OZ2WD446PKRHLRAX","short_pith_number":"pith:AZU3KX4S","schema_version":"1.0","canonical_sha256":"0669b55f9276dd9d587ce79ea89d7105f36e4e14112fcb7430e917af54b05a2e","source":{"kind":"arxiv","id":"2412.19792","version":5},"attestation_state":"computed","paper":{"title":"InfAlign: Inference-aware language model alignment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.IT","math.IT"],"primary_cat":"cs.LG","authors_text":"Adrian Hutter, Ahmad Beirami, Ananda Theertha Suresh, Ananth Balashankar, Aradhana Sinha, Chirag Nagpal, Flavien Prost, Jacob Eisenstein, Jonathan Berant, Jong Lee, Michael Collins, Ziteng Sun","submitted_at":"2024-12-27T18:45:36Z","abstract_excerpt":"Language model alignment is a critical step in training modern generative language models. Alignment targets to improve win rate of a sample from the aligned model against the base model. Today, we are increasingly using inference-time algorithms (e.g., Best-of-N, controlled decoding, tree search) to decode from language models rather than standard sampling. We show that this train/test mismatch makes standard RLHF framework sub-optimal in view of such inference-time methods. To this end, we propose a framework for inference-aware alignment (InfAlign), which aims to optimize inference-time win"},"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":"2412.19792","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-12-27T18:45:36Z","cross_cats_sorted":["cs.CL","cs.IT","math.IT"],"title_canon_sha256":"2fd9580ea20751a67d21081a77fbf10aefa3c1a9502b327e02ac8ddc350add0e","abstract_canon_sha256":"a96742b3582f6292a674bea75d452dc26cc5bacab1f44987ccfeff33ff667c2a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:56:51.906479Z","signature_b64":"9Z9RmYAVciDYgRD/wryrikAkhDDWd8exC23bxSriV47yKgWipA+VswDvhkBSs/5s8pYasIY6NQjPTQO82KlnAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0669b55f9276dd9d587ce79ea89d7105f36e4e14112fcb7430e917af54b05a2e","last_reissued_at":"2026-07-05T11:56:51.905915Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:56:51.905915Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"InfAlign: Inference-aware language model alignment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.IT","math.IT"],"primary_cat":"cs.LG","authors_text":"Adrian Hutter, Ahmad Beirami, Ananda Theertha Suresh, Ananth Balashankar, Aradhana Sinha, Chirag Nagpal, Flavien Prost, Jacob Eisenstein, Jonathan Berant, Jong Lee, Michael Collins, Ziteng Sun","submitted_at":"2024-12-27T18:45:36Z","abstract_excerpt":"Language model alignment is a critical step in training modern generative language models. Alignment targets to improve win rate of a sample from the aligned model against the base model. Today, we are increasingly using inference-time algorithms (e.g., Best-of-N, controlled decoding, tree search) to decode from language models rather than standard sampling. We show that this train/test mismatch makes standard RLHF framework sub-optimal in view of such inference-time methods. To this end, we propose a framework for inference-aware alignment (InfAlign), which aims to optimize inference-time win"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.19792","kind":"arxiv","version":5},"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/2412.19792/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":"2412.19792","created_at":"2026-07-05T11:56:51.905982+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.19792v5","created_at":"2026-07-05T11:56:51.905982+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.19792","created_at":"2026-07-05T11:56:51.905982+00:00"},{"alias_kind":"pith_short_12","alias_value":"AZU3KX4SO3OZ","created_at":"2026-07-05T11:56:51.905982+00:00"},{"alias_kind":"pith_short_16","alias_value":"AZU3KX4SO3OZ2WD4","created_at":"2026-07-05T11:56:51.905982+00:00"},{"alias_kind":"pith_short_8","alias_value":"AZU3KX4S","created_at":"2026-07-05T11:56:51.905982+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.09584","citing_title":"CLR-voyance: Reinforcing Open-Ended Reasoning for Inpatient Clinical Decision Support with Outcome-Aware Rubrics","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10716","citing_title":"What should post-training optimize? A test-time scaling law perspective","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.25872","citing_title":"When Errors Can Be Beneficial: A Categorization of Imperfect Rewards for Policy Gradient","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AZU3KX4SO3OZ2WD446PKRHLRAX","json":"https://pith.science/pith/AZU3KX4SO3OZ2WD446PKRHLRAX.json","graph_json":"https://pith.science/api/pith-number/AZU3KX4SO3OZ2WD446PKRHLRAX/graph.json","events_json":"https://pith.science/api/pith-number/AZU3KX4SO3OZ2WD446PKRHLRAX/events.json","paper":"https://pith.science/paper/AZU3KX4S"},"agent_actions":{"view_html":"https://pith.science/pith/AZU3KX4SO3OZ2WD446PKRHLRAX","download_json":"https://pith.science/pith/AZU3KX4SO3OZ2WD446PKRHLRAX.json","view_paper":"https://pith.science/paper/AZU3KX4S","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.19792&json=true","fetch_graph":"https://pith.science/api/pith-number/AZU3KX4SO3OZ2WD446PKRHLRAX/graph.json","fetch_events":"https://pith.science/api/pith-number/AZU3KX4SO3OZ2WD446PKRHLRAX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AZU3KX4SO3OZ2WD446PKRHLRAX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AZU3KX4SO3OZ2WD446PKRHLRAX/action/storage_attestation","attest_author":"https://pith.science/pith/AZU3KX4SO3OZ2WD446PKRHLRAX/action/author_attestation","sign_citation":"https://pith.science/pith/AZU3KX4SO3OZ2WD446PKRHLRAX/action/citation_signature","submit_replication":"https://pith.science/pith/AZU3KX4SO3OZ2WD446PKRHLRAX/action/replication_record"}},"created_at":"2026-07-05T11:56:51.905982+00:00","updated_at":"2026-07-05T11:56:51.905982+00:00"}