{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:V5TJCKKIQ4ZI5ITRFL73YGNPUQ","short_pith_number":"pith:V5TJCKKI","schema_version":"1.0","canonical_sha256":"af6691294887328ea2712affbc19afa40251c9a1501307a089528a51bc334b57","source":{"kind":"arxiv","id":"2501.04266","version":2},"attestation_state":"computed","paper":{"title":"Scaling Large Language Model Training on Frontier with Low-Bandwidth Partitioning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Aamir Shafi, Dhabaleswar K. (DK) Panda, Hari Subramoni, Jacob Hatef, Lang Xu, Quentin Anthony","submitted_at":"2025-01-08T04:19:57Z","abstract_excerpt":"Scaling up Large Language Model(LLM) training involves fitting a tremendous amount of training parameters across a limited number of workers. However, methods like ZeRO-3 that drastically reduce GPU memory pressure often incur heavy communication to ensure global synchronization and consistency. Established efforts such as ZeRO++ use secondary partitions to avoid inter-node communications, given that intra-node GPU-GPU transfer generally has more bandwidth and lower latency than inter-node connections. However, as more capable infrastructure like Frontier, equipped with AMD GPUs, emerged with "},"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":"2501.04266","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2025-01-08T04:19:57Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"dc0422543c7b698708b542d44ad59323afce8daf81e9fd3987eebac8cd210e1d","abstract_canon_sha256":"1e9cdf2f73b718908712cbaa3af2d5495ccedf74f71db4bac08bc7a34563c524"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:09:14.630332Z","signature_b64":"tejzBmvCvZMgZPa7sDeWf95rxJIJISCYUE0zBzd+/WT2Lm+eRJDPqMxtJB3rtS8CB6cjeHEiPsUhv7dKDRJGBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"af6691294887328ea2712affbc19afa40251c9a1501307a089528a51bc334b57","last_reissued_at":"2026-07-05T10:09:14.629836Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:09:14.629836Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Large Language Model Training on Frontier with Low-Bandwidth Partitioning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Aamir Shafi, Dhabaleswar K. (DK) Panda, Hari Subramoni, Jacob Hatef, Lang Xu, Quentin Anthony","submitted_at":"2025-01-08T04:19:57Z","abstract_excerpt":"Scaling up Large Language Model(LLM) training involves fitting a tremendous amount of training parameters across a limited number of workers. However, methods like ZeRO-3 that drastically reduce GPU memory pressure often incur heavy communication to ensure global synchronization and consistency. Established efforts such as ZeRO++ use secondary partitions to avoid inter-node communications, given that intra-node GPU-GPU transfer generally has more bandwidth and lower latency than inter-node connections. However, as more capable infrastructure like Frontier, equipped with AMD GPUs, emerged with "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.04266","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/2501.04266/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":"2501.04266","created_at":"2026-07-05T10:09:14.629895+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.04266v2","created_at":"2026-07-05T10:09:14.629895+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.04266","created_at":"2026-07-05T10:09:14.629895+00:00"},{"alias_kind":"pith_short_12","alias_value":"V5TJCKKIQ4ZI","created_at":"2026-07-05T10:09:14.629895+00:00"},{"alias_kind":"pith_short_16","alias_value":"V5TJCKKIQ4ZI5ITR","created_at":"2026-07-05T10:09:14.629895+00:00"},{"alias_kind":"pith_short_8","alias_value":"V5TJCKKI","created_at":"2026-07-05T10:09:14.629895+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/V5TJCKKIQ4ZI5ITRFL73YGNPUQ","json":"https://pith.science/pith/V5TJCKKIQ4ZI5ITRFL73YGNPUQ.json","graph_json":"https://pith.science/api/pith-number/V5TJCKKIQ4ZI5ITRFL73YGNPUQ/graph.json","events_json":"https://pith.science/api/pith-number/V5TJCKKIQ4ZI5ITRFL73YGNPUQ/events.json","paper":"https://pith.science/paper/V5TJCKKI"},"agent_actions":{"view_html":"https://pith.science/pith/V5TJCKKIQ4ZI5ITRFL73YGNPUQ","download_json":"https://pith.science/pith/V5TJCKKIQ4ZI5ITRFL73YGNPUQ.json","view_paper":"https://pith.science/paper/V5TJCKKI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.04266&json=true","fetch_graph":"https://pith.science/api/pith-number/V5TJCKKIQ4ZI5ITRFL73YGNPUQ/graph.json","fetch_events":"https://pith.science/api/pith-number/V5TJCKKIQ4ZI5ITRFL73YGNPUQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/V5TJCKKIQ4ZI5ITRFL73YGNPUQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/V5TJCKKIQ4ZI5ITRFL73YGNPUQ/action/storage_attestation","attest_author":"https://pith.science/pith/V5TJCKKIQ4ZI5ITRFL73YGNPUQ/action/author_attestation","sign_citation":"https://pith.science/pith/V5TJCKKIQ4ZI5ITRFL73YGNPUQ/action/citation_signature","submit_replication":"https://pith.science/pith/V5TJCKKIQ4ZI5ITRFL73YGNPUQ/action/replication_record"}},"created_at":"2026-07-05T10:09:14.629895+00:00","updated_at":"2026-07-05T10:09:14.629895+00:00"}