{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:TW4E7NVWLXPOVOXZ5XQHFPVKKN","short_pith_number":"pith:TW4E7NVW","schema_version":"1.0","canonical_sha256":"9db84fb6b65ddeeabaf9ede072beaa5351078a0497ef11bb80521d5190ee0010","source":{"kind":"arxiv","id":"2411.01075","version":2},"attestation_state":"computed","paper":{"title":"Cephalo: Harnessing Heterogeneous GPU Clusters for Training Transformer Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Arthur Chen, Khuzaima Daudjee, Runsheng Benson Guo, Utkarsh Anand","submitted_at":"2024-11-01T23:09:44Z","abstract_excerpt":"Training transformer models requires substantial GPU compute and memory resources. In homogeneous clusters, distributed strategies allocate resources evenly, but this approach is inefficient for heterogeneous clusters, where GPUs differ in power and memory. As high-end GPUs are costly and limited in availability, heterogeneous clusters with diverse GPU types are becoming more common. Existing methods attempt to balance compute across GPUs based on capacity but often underutilize compute due to memory constraints. We present Cephalo, a system that optimizes compute and memory usage by decouplin"},"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":"2411.01075","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.DC","submitted_at":"2024-11-01T23:09:44Z","cross_cats_sorted":[],"title_canon_sha256":"d18902bcadecf4855e5848ededce544e52a454fdd66bed53b8a57c6f6317cf39","abstract_canon_sha256":"30e883588a442434c5e48d69d69bf2dda02903dc052c722339beece8bb7eafe8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:35:14.175640Z","signature_b64":"euignFgKMj+PeK4wk0HuRyyqNyOAWZtwwD5S+RsqYhbdkCVwkGj7Lp2I8tGs3OXtxUCvLIBbUezyj4MpG/S8Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9db84fb6b65ddeeabaf9ede072beaa5351078a0497ef11bb80521d5190ee0010","last_reissued_at":"2026-07-05T09:35:14.175158Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:35:14.175158Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Cephalo: Harnessing Heterogeneous GPU Clusters for Training Transformer Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Arthur Chen, Khuzaima Daudjee, Runsheng Benson Guo, Utkarsh Anand","submitted_at":"2024-11-01T23:09:44Z","abstract_excerpt":"Training transformer models requires substantial GPU compute and memory resources. In homogeneous clusters, distributed strategies allocate resources evenly, but this approach is inefficient for heterogeneous clusters, where GPUs differ in power and memory. As high-end GPUs are costly and limited in availability, heterogeneous clusters with diverse GPU types are becoming more common. Existing methods attempt to balance compute across GPUs based on capacity but often underutilize compute due to memory constraints. We present Cephalo, a system that optimizes compute and memory usage by decouplin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.01075","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/2411.01075/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":"2411.01075","created_at":"2026-07-05T09:35:14.175217+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.01075v2","created_at":"2026-07-05T09:35:14.175217+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.01075","created_at":"2026-07-05T09:35:14.175217+00:00"},{"alias_kind":"pith_short_12","alias_value":"TW4E7NVWLXPO","created_at":"2026-07-05T09:35:14.175217+00:00"},{"alias_kind":"pith_short_16","alias_value":"TW4E7NVWLXPOVOXZ","created_at":"2026-07-05T09:35:14.175217+00:00"},{"alias_kind":"pith_short_8","alias_value":"TW4E7NVW","created_at":"2026-07-05T09:35:14.175217+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26633","citing_title":"Simulating Unified Tensor Resharding in heterogeneous AI systems","ref_index":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TW4E7NVWLXPOVOXZ5XQHFPVKKN","json":"https://pith.science/pith/TW4E7NVWLXPOVOXZ5XQHFPVKKN.json","graph_json":"https://pith.science/api/pith-number/TW4E7NVWLXPOVOXZ5XQHFPVKKN/graph.json","events_json":"https://pith.science/api/pith-number/TW4E7NVWLXPOVOXZ5XQHFPVKKN/events.json","paper":"https://pith.science/paper/TW4E7NVW"},"agent_actions":{"view_html":"https://pith.science/pith/TW4E7NVWLXPOVOXZ5XQHFPVKKN","download_json":"https://pith.science/pith/TW4E7NVWLXPOVOXZ5XQHFPVKKN.json","view_paper":"https://pith.science/paper/TW4E7NVW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.01075&json=true","fetch_graph":"https://pith.science/api/pith-number/TW4E7NVWLXPOVOXZ5XQHFPVKKN/graph.json","fetch_events":"https://pith.science/api/pith-number/TW4E7NVWLXPOVOXZ5XQHFPVKKN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TW4E7NVWLXPOVOXZ5XQHFPVKKN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TW4E7NVWLXPOVOXZ5XQHFPVKKN/action/storage_attestation","attest_author":"https://pith.science/pith/TW4E7NVWLXPOVOXZ5XQHFPVKKN/action/author_attestation","sign_citation":"https://pith.science/pith/TW4E7NVWLXPOVOXZ5XQHFPVKKN/action/citation_signature","submit_replication":"https://pith.science/pith/TW4E7NVWLXPOVOXZ5XQHFPVKKN/action/replication_record"}},"created_at":"2026-07-05T09:35:14.175217+00:00","updated_at":"2026-07-05T09:35:14.175217+00:00"}