{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:XVAUXXEF5ASSIUY3UT7HOGUKB6","short_pith_number":"pith:XVAUXXEF","schema_version":"1.0","canonical_sha256":"bd414bdc85e82524531ba4fe771a8a0f93530b13299284c7ada5bc37469017e2","source":{"kind":"arxiv","id":"2302.11665","version":2},"attestation_state":"computed","paper":{"title":"AlpaServe: Statistical Multiplexing with Model Parallelism for Deep Learning Serving","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.DC","cs.NI"],"primary_cat":"cs.LG","authors_text":"Hao Zhang, Ion Stoica, Joseph E. Gonzalez, Lianmin Zheng, Vincent Liu, Xin Jin, Yanping Huang, Ying Sheng, Yinmin Zhong, Zhifeng Chen, Zhuohan Li","submitted_at":"2023-02-22T21:41:34Z","abstract_excerpt":"Model parallelism is conventionally viewed as a method to scale a single large deep learning model beyond the memory limits of a single device. In this paper, we demonstrate that model parallelism can be additionally used for the statistical multiplexing of multiple devices when serving multiple models, even when a single model can fit into a single device. Our work reveals a fundamental trade-off between the overhead introduced by model parallelism and the opportunity to exploit statistical multiplexing to reduce serving latency in the presence of bursty workloads. We explore the new trade-of"},"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":"2302.11665","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-02-22T21:41:34Z","cross_cats_sorted":["cs.DC","cs.NI"],"title_canon_sha256":"23a6eccbcafdfda41e3e759b1cb62b300fc79db57b9d51bb1d4902a212b90078","abstract_canon_sha256":"b1932fa60a7d6117dded1ade57d5a2c124aa142e353fea75404db309a5901760"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:32:36.912020Z","signature_b64":"e9gq4xeNbGukPuC66GkbtEADKa/xjwJ4bL2zc2VvrleEFGwtXjKZawzjU72t1qn+yfc263324KEpF6JSOBr8CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bd414bdc85e82524531ba4fe771a8a0f93530b13299284c7ada5bc37469017e2","last_reissued_at":"2026-07-05T06:32:36.911587Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:32:36.911587Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AlpaServe: Statistical Multiplexing with Model Parallelism for Deep Learning Serving","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.DC","cs.NI"],"primary_cat":"cs.LG","authors_text":"Hao Zhang, Ion Stoica, Joseph E. Gonzalez, Lianmin Zheng, Vincent Liu, Xin Jin, Yanping Huang, Ying Sheng, Yinmin Zhong, Zhifeng Chen, Zhuohan Li","submitted_at":"2023-02-22T21:41:34Z","abstract_excerpt":"Model parallelism is conventionally viewed as a method to scale a single large deep learning model beyond the memory limits of a single device. In this paper, we demonstrate that model parallelism can be additionally used for the statistical multiplexing of multiple devices when serving multiple models, even when a single model can fit into a single device. Our work reveals a fundamental trade-off between the overhead introduced by model parallelism and the opportunity to exploit statistical multiplexing to reduce serving latency in the presence of bursty workloads. We explore the new trade-of"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.11665","kind":"arxiv","version":2},"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/2302.11665/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":"2302.11665","created_at":"2026-07-05T06:32:36.911644+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.11665v2","created_at":"2026-07-05T06:32:36.911644+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.11665","created_at":"2026-07-05T06:32:36.911644+00:00"},{"alias_kind":"pith_short_12","alias_value":"XVAUXXEF5ASS","created_at":"2026-07-05T06:32:36.911644+00:00"},{"alias_kind":"pith_short_16","alias_value":"XVAUXXEF5ASSIUY3","created_at":"2026-07-05T06:32:36.911644+00:00"},{"alias_kind":"pith_short_8","alias_value":"XVAUXXEF","created_at":"2026-07-05T06:32:36.911644+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01579","citing_title":"OmniPilot: An Uncertainty-Aware LLM Inference Advisor for Heterogeneous GPU Clusters","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00735","citing_title":"ViBE: Co-Optimizing Workload Skew and Hardware Variability for MoE Serving","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2309.06180","citing_title":"Efficient Memory Management for Large Language Model Serving with PagedAttention","ref_index":30,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XVAUXXEF5ASSIUY3UT7HOGUKB6","json":"https://pith.science/pith/XVAUXXEF5ASSIUY3UT7HOGUKB6.json","graph_json":"https://pith.science/api/pith-number/XVAUXXEF5ASSIUY3UT7HOGUKB6/graph.json","events_json":"https://pith.science/api/pith-number/XVAUXXEF5ASSIUY3UT7HOGUKB6/events.json","paper":"https://pith.science/paper/XVAUXXEF"},"agent_actions":{"view_html":"https://pith.science/pith/XVAUXXEF5ASSIUY3UT7HOGUKB6","download_json":"https://pith.science/pith/XVAUXXEF5ASSIUY3UT7HOGUKB6.json","view_paper":"https://pith.science/paper/XVAUXXEF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.11665&json=true","fetch_graph":"https://pith.science/api/pith-number/XVAUXXEF5ASSIUY3UT7HOGUKB6/graph.json","fetch_events":"https://pith.science/api/pith-number/XVAUXXEF5ASSIUY3UT7HOGUKB6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XVAUXXEF5ASSIUY3UT7HOGUKB6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XVAUXXEF5ASSIUY3UT7HOGUKB6/action/storage_attestation","attest_author":"https://pith.science/pith/XVAUXXEF5ASSIUY3UT7HOGUKB6/action/author_attestation","sign_citation":"https://pith.science/pith/XVAUXXEF5ASSIUY3UT7HOGUKB6/action/citation_signature","submit_replication":"https://pith.science/pith/XVAUXXEF5ASSIUY3UT7HOGUKB6/action/replication_record"}},"created_at":"2026-07-05T06:32:36.911644+00:00","updated_at":"2026-07-05T06:32:36.911644+00:00"}