{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:CHJ5KWTGN6DNXVXUMPALR6BAQB","short_pith_number":"pith:CHJ5KWTG","schema_version":"1.0","canonical_sha256":"11d3d55a666f86dbd6f463c0b8f820806d9bdc84a7e41c3ec72c57a984984304","source":{"kind":"arxiv","id":"2310.06003","version":2},"attestation_state":"computed","paper":{"title":"Rethinking Memory and Communication Cost for Efficient Large Language Model Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chan Wu, Fanzhuang Meng, Hanxiao Zhang, Jinjing Huang, Jun Zhou, Lei Liang, Lin Ju, Siyuan Li, Xiaolu Zhang, Youshao Xiao, Zhaoxin Huan","submitted_at":"2023-10-09T15:08:32Z","abstract_excerpt":"Recently, various distributed strategies for large language model training have been proposed. However, these methods provided limited solutions for the trade-off between memory consumption and communication cost. In this paper, we rethink the impact of memory consumption and communication costs on the training speed of large language models, and propose a memory-communication balanced strategy set Partial Redundancy Optimizer (PaRO). PaRO provides comprehensive options which reduces the amount and frequency of inter-group communication with minor memory redundancy by fine-grained sharding str"},"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":"2310.06003","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-09T15:08:32Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"049ede9a0d05d96834a587df751ce4052af7a682e64c357eded4108b9ea08376","abstract_canon_sha256":"231b27a093e251cf55ccc01a8f0df0b971989722e09ecd01a4bd0757e469ad58"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:06:35.745874Z","signature_b64":"GgTeQ3MMR60YBFNxYCYIiO6z037ZRDiSsonlRyv++pUdEEOuBPHjNKiE3dnuE2zCAvaQ4iM+lUaIUNBOj7a+DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"11d3d55a666f86dbd6f463c0b8f820806d9bdc84a7e41c3ec72c57a984984304","last_reissued_at":"2026-07-05T07:06:35.745332Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:06:35.745332Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Rethinking Memory and Communication Cost for Efficient Large Language Model Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chan Wu, Fanzhuang Meng, Hanxiao Zhang, Jinjing Huang, Jun Zhou, Lei Liang, Lin Ju, Siyuan Li, Xiaolu Zhang, Youshao Xiao, Zhaoxin Huan","submitted_at":"2023-10-09T15:08:32Z","abstract_excerpt":"Recently, various distributed strategies for large language model training have been proposed. However, these methods provided limited solutions for the trade-off between memory consumption and communication cost. In this paper, we rethink the impact of memory consumption and communication costs on the training speed of large language models, and propose a memory-communication balanced strategy set Partial Redundancy Optimizer (PaRO). PaRO provides comprehensive options which reduces the amount and frequency of inter-group communication with minor memory redundancy by fine-grained sharding str"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.06003","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/2310.06003/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":"2310.06003","created_at":"2026-07-05T07:06:35.745405+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.06003v2","created_at":"2026-07-05T07:06:35.745405+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.06003","created_at":"2026-07-05T07:06:35.745405+00:00"},{"alias_kind":"pith_short_12","alias_value":"CHJ5KWTGN6DN","created_at":"2026-07-05T07:06:35.745405+00:00"},{"alias_kind":"pith_short_16","alias_value":"CHJ5KWTGN6DNXVXU","created_at":"2026-07-05T07:06:35.745405+00:00"},{"alias_kind":"pith_short_8","alias_value":"CHJ5KWTG","created_at":"2026-07-05T07:06:35.745405+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.04164","citing_title":"Efficient Distributed Optimization under Heavy-Tailed Noise","ref_index":6,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CHJ5KWTGN6DNXVXUMPALR6BAQB","json":"https://pith.science/pith/CHJ5KWTGN6DNXVXUMPALR6BAQB.json","graph_json":"https://pith.science/api/pith-number/CHJ5KWTGN6DNXVXUMPALR6BAQB/graph.json","events_json":"https://pith.science/api/pith-number/CHJ5KWTGN6DNXVXUMPALR6BAQB/events.json","paper":"https://pith.science/paper/CHJ5KWTG"},"agent_actions":{"view_html":"https://pith.science/pith/CHJ5KWTGN6DNXVXUMPALR6BAQB","download_json":"https://pith.science/pith/CHJ5KWTGN6DNXVXUMPALR6BAQB.json","view_paper":"https://pith.science/paper/CHJ5KWTG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.06003&json=true","fetch_graph":"https://pith.science/api/pith-number/CHJ5KWTGN6DNXVXUMPALR6BAQB/graph.json","fetch_events":"https://pith.science/api/pith-number/CHJ5KWTGN6DNXVXUMPALR6BAQB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CHJ5KWTGN6DNXVXUMPALR6BAQB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CHJ5KWTGN6DNXVXUMPALR6BAQB/action/storage_attestation","attest_author":"https://pith.science/pith/CHJ5KWTGN6DNXVXUMPALR6BAQB/action/author_attestation","sign_citation":"https://pith.science/pith/CHJ5KWTGN6DNXVXUMPALR6BAQB/action/citation_signature","submit_replication":"https://pith.science/pith/CHJ5KWTGN6DNXVXUMPALR6BAQB/action/replication_record"}},"created_at":"2026-07-05T07:06:35.745405+00:00","updated_at":"2026-07-05T07:06:35.745405+00:00"}