{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:TALHUMC4ATLU34NWSM5ZQ4KNHW","short_pith_number":"pith:TALHUMC4","schema_version":"1.0","canonical_sha256":"98167a305c04d74df1b6933b98714d3da2dc66e350ca7994c914e3b0f6a5923d","source":{"kind":"arxiv","id":"2507.12704","version":3},"attestation_state":"computed","paper":{"title":"PinFM: Foundation Model for User Activity Sequences at a Billion-scale Visual Discovery Platform","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.LG","authors_text":"Charles Rosenberg, Hanyu Li, Haomiao Li, Jaewon Yang, Jiajing Xu, Kousik Rajesh, Matthew Lawhon, Pong Eksombatchai, Saurabh Vishwas Joshi, Xiangyi Chen, Yi-Ping Hsu, Zelun Wang","submitted_at":"2025-07-17T00:37:59Z","abstract_excerpt":"User activity sequences have emerged as one of the most important signals in recommender systems. We present a foundational model, PinFM, for understanding user activity sequences across multiple applications at a billion-scale visual discovery platform. We pretrain a transformer model with 20B+ parameters using extensive user activity data, then fine-tune it for specific applications, efficiently coupling it with existing models. While this pretraining-and-fine-tuning approach has been popular in other domains, such as Vision and NLP, its application in industrial recommender systems presents"},"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":"2507.12704","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-17T00:37:59Z","cross_cats_sorted":["cs.IR"],"title_canon_sha256":"476e04e5cfa4c04ed9da6a73e32ed969ff37e22c8929d52955dd05c9d5a6c990","abstract_canon_sha256":"c4489e2e99a14e8c560ac0aa3c20d6ab5da0f5d4511022e3fea851447366548a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:56:59.287967Z","signature_b64":"bR2oSaFr935oLDxXtT8m8wCoFqbhVKhxkKgcd74Uu2rB8QIH7gUnTKbIt9XpO2MTHiGFeBecuxNh6S5ahQe/Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"98167a305c04d74df1b6933b98714d3da2dc66e350ca7994c914e3b0f6a5923d","last_reissued_at":"2026-07-05T11:56:59.287472Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:56:59.287472Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PinFM: Foundation Model for User Activity Sequences at a Billion-scale Visual Discovery Platform","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.LG","authors_text":"Charles Rosenberg, Hanyu Li, Haomiao Li, Jaewon Yang, Jiajing Xu, Kousik Rajesh, Matthew Lawhon, Pong Eksombatchai, Saurabh Vishwas Joshi, Xiangyi Chen, Yi-Ping Hsu, Zelun Wang","submitted_at":"2025-07-17T00:37:59Z","abstract_excerpt":"User activity sequences have emerged as one of the most important signals in recommender systems. We present a foundational model, PinFM, for understanding user activity sequences across multiple applications at a billion-scale visual discovery platform. We pretrain a transformer model with 20B+ parameters using extensive user activity data, then fine-tune it for specific applications, efficiently coupling it with existing models. While this pretraining-and-fine-tuning approach has been popular in other domains, such as Vision and NLP, its application in industrial recommender systems presents"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.12704","kind":"arxiv","version":3},"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/2507.12704/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":"2507.12704","created_at":"2026-07-05T11:56:59.287530+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.12704v3","created_at":"2026-07-05T11:56:59.287530+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.12704","created_at":"2026-07-05T11:56:59.287530+00:00"},{"alias_kind":"pith_short_12","alias_value":"TALHUMC4ATLU","created_at":"2026-07-05T11:56:59.287530+00:00"},{"alias_kind":"pith_short_16","alias_value":"TALHUMC4ATLU34NW","created_at":"2026-07-05T11:56:59.287530+00:00"},{"alias_kind":"pith_short_8","alias_value":"TALHUMC4","created_at":"2026-07-05T11:56:59.287530+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2508.14948","citing_title":"Large Foundation Model for Ads Recommendation","ref_index":2025,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TALHUMC4ATLU34NWSM5ZQ4KNHW","json":"https://pith.science/pith/TALHUMC4ATLU34NWSM5ZQ4KNHW.json","graph_json":"https://pith.science/api/pith-number/TALHUMC4ATLU34NWSM5ZQ4KNHW/graph.json","events_json":"https://pith.science/api/pith-number/TALHUMC4ATLU34NWSM5ZQ4KNHW/events.json","paper":"https://pith.science/paper/TALHUMC4"},"agent_actions":{"view_html":"https://pith.science/pith/TALHUMC4ATLU34NWSM5ZQ4KNHW","download_json":"https://pith.science/pith/TALHUMC4ATLU34NWSM5ZQ4KNHW.json","view_paper":"https://pith.science/paper/TALHUMC4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.12704&json=true","fetch_graph":"https://pith.science/api/pith-number/TALHUMC4ATLU34NWSM5ZQ4KNHW/graph.json","fetch_events":"https://pith.science/api/pith-number/TALHUMC4ATLU34NWSM5ZQ4KNHW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TALHUMC4ATLU34NWSM5ZQ4KNHW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TALHUMC4ATLU34NWSM5ZQ4KNHW/action/storage_attestation","attest_author":"https://pith.science/pith/TALHUMC4ATLU34NWSM5ZQ4KNHW/action/author_attestation","sign_citation":"https://pith.science/pith/TALHUMC4ATLU34NWSM5ZQ4KNHW/action/citation_signature","submit_replication":"https://pith.science/pith/TALHUMC4ATLU34NWSM5ZQ4KNHW/action/replication_record"}},"created_at":"2026-07-05T11:56:59.287530+00:00","updated_at":"2026-07-05T11:56:59.287530+00:00"}