{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KZUJ23VRYYQROZVJOG27IW2VGG","short_pith_number":"pith:KZUJ23VR","schema_version":"1.0","canonical_sha256":"56689d6eb1c6211766a971b5f45b5531952d7fbdb10360908f826cc969abba35","source":{"kind":"arxiv","id":"2504.04178","version":4},"attestation_state":"computed","paper":{"title":"MSL: Not All Tokens Are What You Need for Tuning LLM as a Recommender","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Bohao Wang, Can Wang, Changwang Zhang, Chun Chen, Feng Liu, Jiawei Chen, Jun Wang, Xingyu Lou, Yan Feng, Yuegang Sun","submitted_at":"2025-04-05T13:48:33Z","abstract_excerpt":"Large language models (LLMs), known for their comprehension capabilities and extensive knowledge, have been increasingly applied to recommendation systems (RS). Given the fundamental gap between the mechanism of LLMs and the requirement of RS, researchers have focused on fine-tuning LLMs with recommendation-specific data to enhance their performance. Language Modeling Loss (LML), originally designed for language generation tasks, is commonly adopted. However, we identify two critical limitations of LML: 1) it exhibits significant divergence from the recommendation objective; 2) it erroneously "},"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":"2504.04178","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2025-04-05T13:48:33Z","cross_cats_sorted":[],"title_canon_sha256":"2a8101af0599ed17017b1c44f662a9300c6807cea3d86db55216c83dd5f4c935","abstract_canon_sha256":"219b0ca8da9c3603ab99d202a6b906d28b6f4ac9d0e50da4a18084f9392b211f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:17:44.534983Z","signature_b64":"QW8F7Y9Bn/p7L9fR8j17dAnTg7RxnK4sk62jdpeL2jfREU6HQZU0qW7Wanq7AvPKW89m3Q73xZVZnfS6qhaVAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"56689d6eb1c6211766a971b5f45b5531952d7fbdb10360908f826cc969abba35","last_reissued_at":"2026-07-05T11:17:44.534475Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:17:44.534475Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MSL: Not All Tokens Are What You Need for Tuning LLM as a Recommender","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Bohao Wang, Can Wang, Changwang Zhang, Chun Chen, Feng Liu, Jiawei Chen, Jun Wang, Xingyu Lou, Yan Feng, Yuegang Sun","submitted_at":"2025-04-05T13:48:33Z","abstract_excerpt":"Large language models (LLMs), known for their comprehension capabilities and extensive knowledge, have been increasingly applied to recommendation systems (RS). Given the fundamental gap between the mechanism of LLMs and the requirement of RS, researchers have focused on fine-tuning LLMs with recommendation-specific data to enhance their performance. Language Modeling Loss (LML), originally designed for language generation tasks, is commonly adopted. However, we identify two critical limitations of LML: 1) it exhibits significant divergence from the recommendation objective; 2) it erroneously "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.04178","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/2504.04178/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":"2504.04178","created_at":"2026-07-05T11:17:44.534552+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.04178v4","created_at":"2026-07-05T11:17:44.534552+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.04178","created_at":"2026-07-05T11:17:44.534552+00:00"},{"alias_kind":"pith_short_12","alias_value":"KZUJ23VRYYQR","created_at":"2026-07-05T11:17:44.534552+00:00"},{"alias_kind":"pith_short_16","alias_value":"KZUJ23VRYYQROZVJ","created_at":"2026-07-05T11:17:44.534552+00:00"},{"alias_kind":"pith_short_8","alias_value":"KZUJ23VR","created_at":"2026-07-05T11:17:44.534552+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/KZUJ23VRYYQROZVJOG27IW2VGG","json":"https://pith.science/pith/KZUJ23VRYYQROZVJOG27IW2VGG.json","graph_json":"https://pith.science/api/pith-number/KZUJ23VRYYQROZVJOG27IW2VGG/graph.json","events_json":"https://pith.science/api/pith-number/KZUJ23VRYYQROZVJOG27IW2VGG/events.json","paper":"https://pith.science/paper/KZUJ23VR"},"agent_actions":{"view_html":"https://pith.science/pith/KZUJ23VRYYQROZVJOG27IW2VGG","download_json":"https://pith.science/pith/KZUJ23VRYYQROZVJOG27IW2VGG.json","view_paper":"https://pith.science/paper/KZUJ23VR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.04178&json=true","fetch_graph":"https://pith.science/api/pith-number/KZUJ23VRYYQROZVJOG27IW2VGG/graph.json","fetch_events":"https://pith.science/api/pith-number/KZUJ23VRYYQROZVJOG27IW2VGG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KZUJ23VRYYQROZVJOG27IW2VGG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KZUJ23VRYYQROZVJOG27IW2VGG/action/storage_attestation","attest_author":"https://pith.science/pith/KZUJ23VRYYQROZVJOG27IW2VGG/action/author_attestation","sign_citation":"https://pith.science/pith/KZUJ23VRYYQROZVJOG27IW2VGG/action/citation_signature","submit_replication":"https://pith.science/pith/KZUJ23VRYYQROZVJOG27IW2VGG/action/replication_record"}},"created_at":"2026-07-05T11:17:44.534552+00:00","updated_at":"2026-07-05T11:17:44.534552+00:00"}