{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:DMNLJ6Y62I7FAE2FJDFAZ24ALN","short_pith_number":"pith:DMNLJ6Y6","schema_version":"1.0","canonical_sha256":"1b1ab4fb1ed23e50134548ca0ceb805b637dc44de6e826acaa4d6c68ac5aa1c5","source":{"kind":"arxiv","id":"2407.00599","version":2},"attestation_state":"computed","paper":{"title":"Parm: Efficient Training of Large Sparsely-Activated Models with Dedicated Schedules","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Bo Li, Shaohuai Shi, Weinong Sun, Wenxiang Lin, Xiaowen Chu, Xinglin Pan","submitted_at":"2024-06-30T05:55:11Z","abstract_excerpt":"Sparsely-activated Mixture-of-Expert (MoE) layers have found practical applications in enlarging the model size of large-scale foundation models, with only a sub-linear increase in computation demands. Despite the wide adoption of hybrid parallel paradigms like model parallelism, expert parallelism, and expert-sharding parallelism (i.e., MP+EP+ESP) to support MoE model training on GPU clusters, the training efficiency is hindered by communication costs introduced by these parallel paradigms. To address this limitation, we propose Parm, a system that accelerates MP+EP+ESP training by designing "},"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":"2407.00599","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2024-06-30T05:55:11Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"f5c2c209cc2ea0820360a8fff807ffcf0980155cb6e9011ddecca998a4e5bb16","abstract_canon_sha256":"7ec4115563e9a6aff4222a304189546c9dced99181ce34f2f98bc321a095dbd0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:39:36.000609Z","signature_b64":"12LXnA/MBhIOrgTwt5zFD8i7bDpmKnQdmCF/CRGLnD3ORWF/taVJtA64jD+ZcqjE/Yz/MMRLDf2HIokMKU+PBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1b1ab4fb1ed23e50134548ca0ceb805b637dc44de6e826acaa4d6c68ac5aa1c5","last_reissued_at":"2026-07-05T08:39:36.000079Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:39:36.000079Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Parm: Efficient Training of Large Sparsely-Activated Models with Dedicated Schedules","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Bo Li, Shaohuai Shi, Weinong Sun, Wenxiang Lin, Xiaowen Chu, Xinglin Pan","submitted_at":"2024-06-30T05:55:11Z","abstract_excerpt":"Sparsely-activated Mixture-of-Expert (MoE) layers have found practical applications in enlarging the model size of large-scale foundation models, with only a sub-linear increase in computation demands. Despite the wide adoption of hybrid parallel paradigms like model parallelism, expert parallelism, and expert-sharding parallelism (i.e., MP+EP+ESP) to support MoE model training on GPU clusters, the training efficiency is hindered by communication costs introduced by these parallel paradigms. To address this limitation, we propose Parm, a system that accelerates MP+EP+ESP training by designing "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.00599","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/2407.00599/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":"2407.00599","created_at":"2026-07-05T08:39:36.000141+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.00599v2","created_at":"2026-07-05T08:39:36.000141+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.00599","created_at":"2026-07-05T08:39:36.000141+00:00"},{"alias_kind":"pith_short_12","alias_value":"DMNLJ6Y62I7F","created_at":"2026-07-05T08:39:36.000141+00:00"},{"alias_kind":"pith_short_16","alias_value":"DMNLJ6Y62I7FAE2F","created_at":"2026-07-05T08:39:36.000141+00:00"},{"alias_kind":"pith_short_8","alias_value":"DMNLJ6Y6","created_at":"2026-07-05T08:39:36.000141+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/DMNLJ6Y62I7FAE2FJDFAZ24ALN","json":"https://pith.science/pith/DMNLJ6Y62I7FAE2FJDFAZ24ALN.json","graph_json":"https://pith.science/api/pith-number/DMNLJ6Y62I7FAE2FJDFAZ24ALN/graph.json","events_json":"https://pith.science/api/pith-number/DMNLJ6Y62I7FAE2FJDFAZ24ALN/events.json","paper":"https://pith.science/paper/DMNLJ6Y6"},"agent_actions":{"view_html":"https://pith.science/pith/DMNLJ6Y62I7FAE2FJDFAZ24ALN","download_json":"https://pith.science/pith/DMNLJ6Y62I7FAE2FJDFAZ24ALN.json","view_paper":"https://pith.science/paper/DMNLJ6Y6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.00599&json=true","fetch_graph":"https://pith.science/api/pith-number/DMNLJ6Y62I7FAE2FJDFAZ24ALN/graph.json","fetch_events":"https://pith.science/api/pith-number/DMNLJ6Y62I7FAE2FJDFAZ24ALN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DMNLJ6Y62I7FAE2FJDFAZ24ALN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DMNLJ6Y62I7FAE2FJDFAZ24ALN/action/storage_attestation","attest_author":"https://pith.science/pith/DMNLJ6Y62I7FAE2FJDFAZ24ALN/action/author_attestation","sign_citation":"https://pith.science/pith/DMNLJ6Y62I7FAE2FJDFAZ24ALN/action/citation_signature","submit_replication":"https://pith.science/pith/DMNLJ6Y62I7FAE2FJDFAZ24ALN/action/replication_record"}},"created_at":"2026-07-05T08:39:36.000141+00:00","updated_at":"2026-07-05T08:39:36.000141+00:00"}