{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4F47WZKMWXPUQ7DX6KAARP5YKA","short_pith_number":"pith:4F47WZKM","schema_version":"1.0","canonical_sha256":"e179fb654cb5df487c77f28008bfb8503dc76381bc22abfa9516d12ed2c5dca1","source":{"kind":"arxiv","id":"2411.00284","version":2},"attestation_state":"computed","paper":{"title":"SimpleFSDP: Simpler Fully Sharded Data Parallel with torch.compile","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Andrew Gu, Francisco Massa, Ruisi Zhang, Sanket Purandare, Tianyu Liu, Wanchao Liang, Will Feng","submitted_at":"2024-11-01T00:43:54Z","abstract_excerpt":"Distributed training of large models consumes enormous computation resources and requires substantial engineering efforts to compose various training techniques. This paper presents SimpleFSDP, a PyTorch-native compiler-based Fully Sharded Data Parallel (FSDP) framework, which has a simple implementation for maintenance and composability, allows full computation-communication graph tracing, and brings performance enhancement via compiler backend optimizations.\n  SimpleFSDP's novelty lies in its unique $torch.compile$-friendly implementation of collective communications using existing PyTorch p"},"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":"2411.00284","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2024-11-01T00:43:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"a830a18e9c572f4edc286bde086bab505e4881c52ff3e410b12ee512b911c1c3","abstract_canon_sha256":"d91ca5984616eebdeef5559304d4877d58d6b291748f3c2768b84ba2feebbec8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:31:31.060848Z","signature_b64":"R8Z18/eaEgTD1fJgU4mRhzZ3O7jlg+Ram25dqtAdAcPeFNr+9glXm9cf6ZZMNRkno3kXfEQUE+16vh+flbLGDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e179fb654cb5df487c77f28008bfb8503dc76381bc22abfa9516d12ed2c5dca1","last_reissued_at":"2026-07-05T09:31:31.060358Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:31:31.060358Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SimpleFSDP: Simpler Fully Sharded Data Parallel with torch.compile","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Andrew Gu, Francisco Massa, Ruisi Zhang, Sanket Purandare, Tianyu Liu, Wanchao Liang, Will Feng","submitted_at":"2024-11-01T00:43:54Z","abstract_excerpt":"Distributed training of large models consumes enormous computation resources and requires substantial engineering efforts to compose various training techniques. This paper presents SimpleFSDP, a PyTorch-native compiler-based Fully Sharded Data Parallel (FSDP) framework, which has a simple implementation for maintenance and composability, allows full computation-communication graph tracing, and brings performance enhancement via compiler backend optimizations.\n  SimpleFSDP's novelty lies in its unique $torch.compile$-friendly implementation of collective communications using existing PyTorch p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.00284","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/2411.00284/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":"2411.00284","created_at":"2026-07-05T09:31:31.060415+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.00284v2","created_at":"2026-07-05T09:31:31.060415+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.00284","created_at":"2026-07-05T09:31:31.060415+00:00"},{"alias_kind":"pith_short_12","alias_value":"4F47WZKMWXPU","created_at":"2026-07-05T09:31:31.060415+00:00"},{"alias_kind":"pith_short_16","alias_value":"4F47WZKMWXPUQ7DX","created_at":"2026-07-05T09:31:31.060415+00:00"},{"alias_kind":"pith_short_8","alias_value":"4F47WZKM","created_at":"2026-07-05T09:31:31.060415+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.17550","citing_title":"Flint: Compiler Enabled Cluster-Free Design Space Exploration for Distributed ML","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4F47WZKMWXPUQ7DX6KAARP5YKA","json":"https://pith.science/pith/4F47WZKMWXPUQ7DX6KAARP5YKA.json","graph_json":"https://pith.science/api/pith-number/4F47WZKMWXPUQ7DX6KAARP5YKA/graph.json","events_json":"https://pith.science/api/pith-number/4F47WZKMWXPUQ7DX6KAARP5YKA/events.json","paper":"https://pith.science/paper/4F47WZKM"},"agent_actions":{"view_html":"https://pith.science/pith/4F47WZKMWXPUQ7DX6KAARP5YKA","download_json":"https://pith.science/pith/4F47WZKMWXPUQ7DX6KAARP5YKA.json","view_paper":"https://pith.science/paper/4F47WZKM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.00284&json=true","fetch_graph":"https://pith.science/api/pith-number/4F47WZKMWXPUQ7DX6KAARP5YKA/graph.json","fetch_events":"https://pith.science/api/pith-number/4F47WZKMWXPUQ7DX6KAARP5YKA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4F47WZKMWXPUQ7DX6KAARP5YKA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4F47WZKMWXPUQ7DX6KAARP5YKA/action/storage_attestation","attest_author":"https://pith.science/pith/4F47WZKMWXPUQ7DX6KAARP5YKA/action/author_attestation","sign_citation":"https://pith.science/pith/4F47WZKMWXPUQ7DX6KAARP5YKA/action/citation_signature","submit_replication":"https://pith.science/pith/4F47WZKMWXPUQ7DX6KAARP5YKA/action/replication_record"}},"created_at":"2026-07-05T09:31:31.060415+00:00","updated_at":"2026-07-05T09:31:31.060415+00:00"}