{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:AO7GWH7XFYCK5QTESN22BYQL3S","short_pith_number":"pith:AO7GWH7X","schema_version":"1.0","canonical_sha256":"03be6b1ff72e04aec2649375a0e20bdcbaf146d596fddbb948a8c55ba58b3876","source":{"kind":"arxiv","id":"2607.02818","version":1},"attestation_state":"computed","paper":{"title":"Long-Term Optimization for Large-Scale Generative Retrieval with Off-Policy REINFORCE","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Aleksei Krasilnikov, Artem Matveev, Kirill Khrylchenko, Sergei Liamaev, Sergei Makeev, Vladimir Baikalov","submitted_at":"2026-07-02T23:13:36Z","abstract_excerpt":"Generative retrieval has become a popular paradigm for large-scale recommendation. However, it is typically trained with supervised next-item prediction objectives that do not directly optimize long-term user satisfaction.\n  In this work, we formulate recommendation as a session-level sequential decision-making problem and introduce an autoregressive approach for training generative retrievers with off-policy REINFORCE on pre-collected data. Unlike the one-step off-policy correction used in prior work, we propose a multi-step approximation of importance weights enabled by the autoregressive fo"},"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":"2607.02818","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IR","submitted_at":"2026-07-02T23:13:36Z","cross_cats_sorted":[],"title_canon_sha256":"614d841a11822e95b9acf6e89052d5502022e06160257009f8f3586158cf923e","abstract_canon_sha256":"1d888623f62dd7181e8a5b2d97eb31d02d7bb04cb92197932256d6a5e54d78ff"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T00:16:18.975564Z","signature_b64":"pZMPVFV2pT4wadH/kEn5XKAcs/fkYaW8GnYqvrWbpk/hzcURbqKLhWTF6TKdpfPAdeA7e0Hnv4JWExlXxaBjCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"03be6b1ff72e04aec2649375a0e20bdcbaf146d596fddbb948a8c55ba58b3876","last_reissued_at":"2026-07-07T00:16:18.974791Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T00:16:18.974791Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Long-Term Optimization for Large-Scale Generative Retrieval with Off-Policy REINFORCE","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Aleksei Krasilnikov, Artem Matveev, Kirill Khrylchenko, Sergei Liamaev, Sergei Makeev, Vladimir Baikalov","submitted_at":"2026-07-02T23:13:36Z","abstract_excerpt":"Generative retrieval has become a popular paradigm for large-scale recommendation. However, it is typically trained with supervised next-item prediction objectives that do not directly optimize long-term user satisfaction.\n  In this work, we formulate recommendation as a session-level sequential decision-making problem and introduce an autoregressive approach for training generative retrievers with off-policy REINFORCE on pre-collected data. Unlike the one-step off-policy correction used in prior work, we propose a multi-step approximation of importance weights enabled by the autoregressive fo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.02818","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/2607.02818/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":"2607.02818","created_at":"2026-07-07T00:16:18.974907+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.02818v1","created_at":"2026-07-07T00:16:18.974907+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.02818","created_at":"2026-07-07T00:16:18.974907+00:00"},{"alias_kind":"pith_short_12","alias_value":"AO7GWH7XFYCK","created_at":"2026-07-07T00:16:18.974907+00:00"},{"alias_kind":"pith_short_16","alias_value":"AO7GWH7XFYCK5QTE","created_at":"2026-07-07T00:16:18.974907+00:00"},{"alias_kind":"pith_short_8","alias_value":"AO7GWH7X","created_at":"2026-07-07T00:16:18.974907+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/AO7GWH7XFYCK5QTESN22BYQL3S","json":"https://pith.science/pith/AO7GWH7XFYCK5QTESN22BYQL3S.json","graph_json":"https://pith.science/api/pith-number/AO7GWH7XFYCK5QTESN22BYQL3S/graph.json","events_json":"https://pith.science/api/pith-number/AO7GWH7XFYCK5QTESN22BYQL3S/events.json","paper":"https://pith.science/paper/AO7GWH7X"},"agent_actions":{"view_html":"https://pith.science/pith/AO7GWH7XFYCK5QTESN22BYQL3S","download_json":"https://pith.science/pith/AO7GWH7XFYCK5QTESN22BYQL3S.json","view_paper":"https://pith.science/paper/AO7GWH7X","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.02818&json=true","fetch_graph":"https://pith.science/api/pith-number/AO7GWH7XFYCK5QTESN22BYQL3S/graph.json","fetch_events":"https://pith.science/api/pith-number/AO7GWH7XFYCK5QTESN22BYQL3S/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AO7GWH7XFYCK5QTESN22BYQL3S/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AO7GWH7XFYCK5QTESN22BYQL3S/action/storage_attestation","attest_author":"https://pith.science/pith/AO7GWH7XFYCK5QTESN22BYQL3S/action/author_attestation","sign_citation":"https://pith.science/pith/AO7GWH7XFYCK5QTESN22BYQL3S/action/citation_signature","submit_replication":"https://pith.science/pith/AO7GWH7XFYCK5QTESN22BYQL3S/action/replication_record"}},"created_at":"2026-07-07T00:16:18.974907+00:00","updated_at":"2026-07-07T00:16:18.974907+00:00"}