{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HY7URAXVX33PYOLVLI3IOLVCS2","short_pith_number":"pith:HY7URAXV","schema_version":"1.0","canonical_sha256":"3e3f4882f5bef6fc39755a36872ea29688a7f82ce947ec97e77d3476e01b7035","source":{"kind":"arxiv","id":"2608.00816","version":1},"attestation_state":"computed","paper":{"title":"Exponential Reward Weighting for Fine-Tuning Generative Recommenders under Sparse and Noisy Feedback","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Keertana Chidambaram, Ko-Jen Hsiao, Moumita Bhattacharya, Qiuling Xu, Sanath Kumar Krishnamurthy","submitted_at":"2026-08-01T18:48:39Z","abstract_excerpt":"In recommendation systems, users interact with only a small fraction of a vast item catalog, producing feedback that is both sparse and noisy. This challenges post-training generative recommenders: reward models trained from logged interactions often fail to generalize, while directly optimizing imperfect rewards can lead to reward over-optimization. We propose Exponential reward-weighted fine-tuning (Exp-RSFT), where each logged interaction is weighted by $\\exp(r/\\lambda)$, avoids this failure by optimizing directly on the logged rewards, with the temperature $\\lambda$ regularizing against 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":"2608.00816","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.IR","submitted_at":"2026-08-01T18:48:39Z","cross_cats_sorted":[],"title_canon_sha256":"1af7f9cbb3b12605f3110525a380b8b1fc2dd89e6c1b1bcc817379d399be7b94","abstract_canon_sha256":"bb20732a5668eb5e64eca1a14dd73310b8a16d88f9e38e53c6f797ac3cfcee3d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-04T01:55:11.863382Z","signature_b64":"HjwG1Z3E/Ffn7Q8j2wlBjFl8AkoyixxeRcOL4thdUuzqjOVUJVM/SkeuhkBA237nCnNusYy95XQBzZGUIWVeBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3e3f4882f5bef6fc39755a36872ea29688a7f82ce947ec97e77d3476e01b7035","last_reissued_at":"2026-08-04T01:55:11.861765Z","signature_status":"signed_v1","first_computed_at":"2026-08-04T01:55:11.861765Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Exponential Reward Weighting for Fine-Tuning Generative Recommenders under Sparse and Noisy Feedback","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Keertana Chidambaram, Ko-Jen Hsiao, Moumita Bhattacharya, Qiuling Xu, Sanath Kumar Krishnamurthy","submitted_at":"2026-08-01T18:48:39Z","abstract_excerpt":"In recommendation systems, users interact with only a small fraction of a vast item catalog, producing feedback that is both sparse and noisy. This challenges post-training generative recommenders: reward models trained from logged interactions often fail to generalize, while directly optimizing imperfect rewards can lead to reward over-optimization. We propose Exponential reward-weighted fine-tuning (Exp-RSFT), where each logged interaction is weighted by $\\exp(r/\\lambda)$, avoids this failure by optimizing directly on the logged rewards, with the temperature $\\lambda$ regularizing against th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.00816","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/2608.00816/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":"2608.00816","created_at":"2026-08-04T01:55:11.863016+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.00816v1","created_at":"2026-08-04T01:55:11.863016+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.00816","created_at":"2026-08-04T01:55:11.863016+00:00"},{"alias_kind":"pith_short_12","alias_value":"HY7URAXVX33P","created_at":"2026-08-04T01:55:11.863016+00:00"},{"alias_kind":"pith_short_16","alias_value":"HY7URAXVX33PYOLV","created_at":"2026-08-04T01:55:11.863016+00:00"},{"alias_kind":"pith_short_8","alias_value":"HY7URAXV","created_at":"2026-08-04T01:55:11.863016+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/HY7URAXVX33PYOLVLI3IOLVCS2","json":"https://pith.science/pith/HY7URAXVX33PYOLVLI3IOLVCS2.json","graph_json":"https://pith.science/api/pith-number/HY7URAXVX33PYOLVLI3IOLVCS2/graph.json","events_json":"https://pith.science/api/pith-number/HY7URAXVX33PYOLVLI3IOLVCS2/events.json","paper":"https://pith.science/paper/HY7URAXV"},"agent_actions":{"view_html":"https://pith.science/pith/HY7URAXVX33PYOLVLI3IOLVCS2","download_json":"https://pith.science/pith/HY7URAXVX33PYOLVLI3IOLVCS2.json","view_paper":"https://pith.science/paper/HY7URAXV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.00816&json=true","fetch_graph":"https://pith.science/api/pith-number/HY7URAXVX33PYOLVLI3IOLVCS2/graph.json","fetch_events":"https://pith.science/api/pith-number/HY7URAXVX33PYOLVLI3IOLVCS2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HY7URAXVX33PYOLVLI3IOLVCS2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HY7URAXVX33PYOLVLI3IOLVCS2/action/storage_attestation","attest_author":"https://pith.science/pith/HY7URAXVX33PYOLVLI3IOLVCS2/action/author_attestation","sign_citation":"https://pith.science/pith/HY7URAXVX33PYOLVLI3IOLVCS2/action/citation_signature","submit_replication":"https://pith.science/pith/HY7URAXVX33PYOLVLI3IOLVCS2/action/replication_record"}},"created_at":"2026-08-04T01:55:11.863016+00:00","updated_at":"2026-08-04T01:55:11.863016+00:00"}