{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BMWHAH2DRYP4HKSVZEJ4AAHEZE","short_pith_number":"pith:BMWHAH2D","schema_version":"1.0","canonical_sha256":"0b2c701f438e1fc3aa55c913c000e4c92e23fc581826ba745dc5aabde2323b41","source":{"kind":"arxiv","id":"2403.07648","version":2},"attestation_state":"computed","paper":{"title":"Characterization of Large Language Model Development in the Datacenter","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Dahua Lin, Guoteng Wang, Meng Zhang, Peng Sun, Qiaoling Chen, Qinghao Hu, Tianwei Zhang, Xiaolin Wang, Yingwei Luo, Yonggang Wen, Zerui Wang, Zhisheng Ye","submitted_at":"2024-03-12T13:31:14Z","abstract_excerpt":"Large Language Models (LLMs) have presented impressive performance across several transformative tasks. However, it is non-trivial to efficiently utilize large-scale cluster resources to develop LLMs, often riddled with numerous challenges such as frequent hardware failures, intricate parallelization strategies, and imbalanced resource utilization. In this paper, we present an in-depth characterization study of a six-month LLM development workload trace collected from our GPU datacenter Acme. Specifically, we investigate discrepancies between LLMs and prior task-specific Deep Learning (DL) wor"},"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":"2403.07648","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2024-03-12T13:31:14Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"e0d46ea868734cc27b0ecd3de2e6cb6482cb5ee5a819d5e559f00a086596be13","abstract_canon_sha256":"a15fefef2894a8ed1c540c5e4276acd6cbe942e7edcb1bbf5d19641fbc9ecf19"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:04:18.798449Z","signature_b64":"IGWhu1V6tyszOv8M36xY+1/HQQZ905oMyx1o4QLxn4qpYnJ5Q5zgZfuc4HqWtaj0T4wggNbFeIateeHl/Cg9Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0b2c701f438e1fc3aa55c913c000e4c92e23fc581826ba745dc5aabde2323b41","last_reissued_at":"2026-07-05T08:04:18.797964Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:04:18.797964Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Characterization of Large Language Model Development in the Datacenter","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Dahua Lin, Guoteng Wang, Meng Zhang, Peng Sun, Qiaoling Chen, Qinghao Hu, Tianwei Zhang, Xiaolin Wang, Yingwei Luo, Yonggang Wen, Zerui Wang, Zhisheng Ye","submitted_at":"2024-03-12T13:31:14Z","abstract_excerpt":"Large Language Models (LLMs) have presented impressive performance across several transformative tasks. However, it is non-trivial to efficiently utilize large-scale cluster resources to develop LLMs, often riddled with numerous challenges such as frequent hardware failures, intricate parallelization strategies, and imbalanced resource utilization. In this paper, we present an in-depth characterization study of a six-month LLM development workload trace collected from our GPU datacenter Acme. Specifically, we investigate discrepancies between LLMs and prior task-specific Deep Learning (DL) wor"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.07648","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/2403.07648/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":"2403.07648","created_at":"2026-07-05T08:04:18.798018+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.07648v2","created_at":"2026-07-05T08:04:18.798018+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.07648","created_at":"2026-07-05T08:04:18.798018+00:00"},{"alias_kind":"pith_short_12","alias_value":"BMWHAH2DRYP4","created_at":"2026-07-05T08:04:18.798018+00:00"},{"alias_kind":"pith_short_16","alias_value":"BMWHAH2DRYP4HKSV","created_at":"2026-07-05T08:04:18.798018+00:00"},{"alias_kind":"pith_short_8","alias_value":"BMWHAH2D","created_at":"2026-07-05T08:04:18.798018+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2504.09844","citing_title":"MegaScale-Data: Scaling Dataloader for Multisource Large Foundation Model Training","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BMWHAH2DRYP4HKSVZEJ4AAHEZE","json":"https://pith.science/pith/BMWHAH2DRYP4HKSVZEJ4AAHEZE.json","graph_json":"https://pith.science/api/pith-number/BMWHAH2DRYP4HKSVZEJ4AAHEZE/graph.json","events_json":"https://pith.science/api/pith-number/BMWHAH2DRYP4HKSVZEJ4AAHEZE/events.json","paper":"https://pith.science/paper/BMWHAH2D"},"agent_actions":{"view_html":"https://pith.science/pith/BMWHAH2DRYP4HKSVZEJ4AAHEZE","download_json":"https://pith.science/pith/BMWHAH2DRYP4HKSVZEJ4AAHEZE.json","view_paper":"https://pith.science/paper/BMWHAH2D","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.07648&json=true","fetch_graph":"https://pith.science/api/pith-number/BMWHAH2DRYP4HKSVZEJ4AAHEZE/graph.json","fetch_events":"https://pith.science/api/pith-number/BMWHAH2DRYP4HKSVZEJ4AAHEZE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BMWHAH2DRYP4HKSVZEJ4AAHEZE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BMWHAH2DRYP4HKSVZEJ4AAHEZE/action/storage_attestation","attest_author":"https://pith.science/pith/BMWHAH2DRYP4HKSVZEJ4AAHEZE/action/author_attestation","sign_citation":"https://pith.science/pith/BMWHAH2DRYP4HKSVZEJ4AAHEZE/action/citation_signature","submit_replication":"https://pith.science/pith/BMWHAH2DRYP4HKSVZEJ4AAHEZE/action/replication_record"}},"created_at":"2026-07-05T08:04:18.798018+00:00","updated_at":"2026-07-05T08:04:18.798018+00:00"}