{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZDSE6WYLSKKSQOCIQE2LVEBWWW","short_pith_number":"pith:ZDSE6WYL","schema_version":"1.0","canonical_sha256":"c8e44f5b0b92952838488134ba9036b5a59ebfded34a757f613a5b75689f8139","source":{"kind":"arxiv","id":"2507.15551","version":3},"attestation_state":"computed","paper":{"title":"RankMixer: Scaling Up Ranking Models in Industrial Recommenders","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Di Wu, Feng Zhang, Hangyu Wang, Haoran Ding, Huizhi Yang, Jie Zhu, Peng Xu, Qiwei Chen, Wenlin Zhao, Xiaoxie Zhu, Xiao Yang, Xinmin Wang, Xintian Han, Xun Zhou, Yuchao Zheng, Yuchen Jiang, Zhe Chen, Zheng Chai, Zhen Gong, Zhifang Fan, Zuotao Liu","submitted_at":"2025-07-21T12:28:55Z","abstract_excerpt":"Recent progress on large language models (LLMs) has spurred interest in scaling up recommendation systems, yet two practical obstacles remain. First, training and serving cost on industrial Recommenders must respect strict latency bounds and high QPS demands. Second, most human-designed feature-crossing modules in ranking models were inherited from the CPU era and fail to exploit modern GPUs, resulting in low Model Flops Utilization (MFU) and poor scalability. We introduce RankMixer, a hardware-aware model design tailored towards a unified and scalable feature-interaction architecture. RankMix"},"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":"2507.15551","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2025-07-21T12:28:55Z","cross_cats_sorted":[],"title_canon_sha256":"b07ef471397de9f38b08430bd1314b422ff0089b997e88c71f0316ea425232f2","abstract_canon_sha256":"e50aef7722522619d1c49e289f5e16beefc201a38a10bf1bb573cd7370ad49af"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:43:33.932793Z","signature_b64":"LWCiCP5m+VqOjWGoGm9x4RdFC3j81f7u9kc97Cg4m/tYkudGslTgZWeJ7D5mHU1OxawOMI36ue2r9SKg3MVrAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c8e44f5b0b92952838488134ba9036b5a59ebfded34a757f613a5b75689f8139","last_reissued_at":"2026-07-05T11:43:33.932394Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:43:33.932394Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RankMixer: Scaling Up Ranking Models in Industrial Recommenders","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Di Wu, Feng Zhang, Hangyu Wang, Haoran Ding, Huizhi Yang, Jie Zhu, Peng Xu, Qiwei Chen, Wenlin Zhao, Xiaoxie Zhu, Xiao Yang, Xinmin Wang, Xintian Han, Xun Zhou, Yuchao Zheng, Yuchen Jiang, Zhe Chen, Zheng Chai, Zhen Gong, Zhifang Fan, Zuotao Liu","submitted_at":"2025-07-21T12:28:55Z","abstract_excerpt":"Recent progress on large language models (LLMs) has spurred interest in scaling up recommendation systems, yet two practical obstacles remain. First, training and serving cost on industrial Recommenders must respect strict latency bounds and high QPS demands. Second, most human-designed feature-crossing modules in ranking models were inherited from the CPU era and fail to exploit modern GPUs, resulting in low Model Flops Utilization (MFU) and poor scalability. We introduce RankMixer, a hardware-aware model design tailored towards a unified and scalable feature-interaction architecture. RankMix"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.15551","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/2507.15551/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":"2507.15551","created_at":"2026-07-05T11:43:33.932452+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.15551v3","created_at":"2026-07-05T11:43:33.932452+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.15551","created_at":"2026-07-05T11:43:33.932452+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZDSE6WYLSKKS","created_at":"2026-07-05T11:43:33.932452+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZDSE6WYLSKKSQOCI","created_at":"2026-07-05T11:43:33.932452+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZDSE6WYL","created_at":"2026-07-05T11:43:33.932452+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.24989","citing_title":"Selective Test-Time Compute Scaling for Click-Through Rate Prediction via Uncertainty-Triggered Feature Path Exploration","ref_index":63,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29232","citing_title":"On the Practice of Scaling Search Conversion Rate Prediction","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2603.24226","citing_title":"Joint Model Parameter Scaling and Universal-Domain Data Integration for E-commerce Search Ranking","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2511.06077","citing_title":"Make It Long, Keep It Fast: End-to-End 10K Long User Behavior Sequence Modeling for Billion-Scale Douyin Recommendation","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2510.27157","citing_title":"A Survey on Generative Recommendation: Data, Model, and Tasks","ref_index":248,"is_internal_anchor":false},{"citing_arxiv_id":"2511.06077","citing_title":"Make It Long, Keep It Fast: End-to-End 10K Long User Behavior Sequence Modeling for Billion-Scale Douyin Recommendation","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2603.24422","citing_title":"OneSearch-V2: The Latent Reasoning Enhanced Self-distillation Generative Search Framework","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04976","citing_title":"Tencent Advertising Algorithm Challenge 2025: All-Modality Generative Recommendation","ref_index":78,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13737","citing_title":"TokenFormer: Unify the Multi-Field and Sequential Recommendation Worlds","ref_index":61,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZDSE6WYLSKKSQOCIQE2LVEBWWW","json":"https://pith.science/pith/ZDSE6WYLSKKSQOCIQE2LVEBWWW.json","graph_json":"https://pith.science/api/pith-number/ZDSE6WYLSKKSQOCIQE2LVEBWWW/graph.json","events_json":"https://pith.science/api/pith-number/ZDSE6WYLSKKSQOCIQE2LVEBWWW/events.json","paper":"https://pith.science/paper/ZDSE6WYL"},"agent_actions":{"view_html":"https://pith.science/pith/ZDSE6WYLSKKSQOCIQE2LVEBWWW","download_json":"https://pith.science/pith/ZDSE6WYLSKKSQOCIQE2LVEBWWW.json","view_paper":"https://pith.science/paper/ZDSE6WYL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.15551&json=true","fetch_graph":"https://pith.science/api/pith-number/ZDSE6WYLSKKSQOCIQE2LVEBWWW/graph.json","fetch_events":"https://pith.science/api/pith-number/ZDSE6WYLSKKSQOCIQE2LVEBWWW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZDSE6WYLSKKSQOCIQE2LVEBWWW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZDSE6WYLSKKSQOCIQE2LVEBWWW/action/storage_attestation","attest_author":"https://pith.science/pith/ZDSE6WYLSKKSQOCIQE2LVEBWWW/action/author_attestation","sign_citation":"https://pith.science/pith/ZDSE6WYLSKKSQOCIQE2LVEBWWW/action/citation_signature","submit_replication":"https://pith.science/pith/ZDSE6WYLSKKSQOCIQE2LVEBWWW/action/replication_record"}},"created_at":"2026-07-05T11:43:33.932452+00:00","updated_at":"2026-07-05T11:43:33.932452+00:00"}