{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4MHR3EQTPWOTF5RB3ZJXFDAFNN","short_pith_number":"pith:4MHR3EQT","schema_version":"1.0","canonical_sha256":"e30f1d92137d9d32f621de53728c056b69376ad2c8f37f6a40008a823f3e6550","source":{"kind":"arxiv","id":"2411.08446","version":1},"attestation_state":"computed","paper":{"title":"LSH-MoE: Communication-efficient MoE Training via Locality-Sensitive Hashing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Bin Cui, Fangcheng Fu, Qibin Liu, Shenhan Zhu, Shouda Liu, Xiaonan Nie, Xiaoyang Li, Xupeng Miao, Yang Zhang","submitted_at":"2024-11-13T08:57:59Z","abstract_excerpt":"Larger transformer models always perform better on various tasks but require more costs to scale up the model size. To efficiently enlarge models, the mixture-of-experts (MoE) architecture is widely adopted, which consists of a gate network and a series of experts and keep the training cost constant by routing the input data to a fixed number of experts instead of all. In existing large-scale MoE training systems, experts would be distributed among different GPUs for parallelization, and thus input data requires additional all-to-all communications to access the target experts and conduct corr"},"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":"2411.08446","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2024-11-13T08:57:59Z","cross_cats_sorted":[],"title_canon_sha256":"575c2f122f13f9be7efce06ba990bc15bb4e01f2bdd20d021f812d70d2618526","abstract_canon_sha256":"d8942bb62711ddf7f0cbf992f8d6a7dd9ea8c51278f693948623ea2eb0dda8e9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:34:47.755964Z","signature_b64":"KtHMmBu6e/Fp/UfonrKMxPwKJyDdgWq4Qg2JDAZw+buEZOzX1p3XntwVkq8YEFHGlFLqC1abNWaQVPI7+q87AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e30f1d92137d9d32f621de53728c056b69376ad2c8f37f6a40008a823f3e6550","last_reissued_at":"2026-07-05T09:34:47.755486Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:34:47.755486Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LSH-MoE: Communication-efficient MoE Training via Locality-Sensitive Hashing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Bin Cui, Fangcheng Fu, Qibin Liu, Shenhan Zhu, Shouda Liu, Xiaonan Nie, Xiaoyang Li, Xupeng Miao, Yang Zhang","submitted_at":"2024-11-13T08:57:59Z","abstract_excerpt":"Larger transformer models always perform better on various tasks but require more costs to scale up the model size. To efficiently enlarge models, the mixture-of-experts (MoE) architecture is widely adopted, which consists of a gate network and a series of experts and keep the training cost constant by routing the input data to a fixed number of experts instead of all. In existing large-scale MoE training systems, experts would be distributed among different GPUs for parallelization, and thus input data requires additional all-to-all communications to access the target experts and conduct corr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.08446","kind":"arxiv","version":1},"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/2411.08446/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":"2411.08446","created_at":"2026-07-05T09:34:47.755551+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.08446v1","created_at":"2026-07-05T09:34:47.755551+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.08446","created_at":"2026-07-05T09:34:47.755551+00:00"},{"alias_kind":"pith_short_12","alias_value":"4MHR3EQTPWOT","created_at":"2026-07-05T09:34:47.755551+00:00"},{"alias_kind":"pith_short_16","alias_value":"4MHR3EQTPWOTF5RB","created_at":"2026-07-05T09:34:47.755551+00:00"},{"alias_kind":"pith_short_8","alias_value":"4MHR3EQT","created_at":"2026-07-05T09:34:47.755551+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.02523","citing_title":"Brief analysis of DeepSeek R1 and its implications for Generative AI","ref_index":18,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4MHR3EQTPWOTF5RB3ZJXFDAFNN","json":"https://pith.science/pith/4MHR3EQTPWOTF5RB3ZJXFDAFNN.json","graph_json":"https://pith.science/api/pith-number/4MHR3EQTPWOTF5RB3ZJXFDAFNN/graph.json","events_json":"https://pith.science/api/pith-number/4MHR3EQTPWOTF5RB3ZJXFDAFNN/events.json","paper":"https://pith.science/paper/4MHR3EQT"},"agent_actions":{"view_html":"https://pith.science/pith/4MHR3EQTPWOTF5RB3ZJXFDAFNN","download_json":"https://pith.science/pith/4MHR3EQTPWOTF5RB3ZJXFDAFNN.json","view_paper":"https://pith.science/paper/4MHR3EQT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.08446&json=true","fetch_graph":"https://pith.science/api/pith-number/4MHR3EQTPWOTF5RB3ZJXFDAFNN/graph.json","fetch_events":"https://pith.science/api/pith-number/4MHR3EQTPWOTF5RB3ZJXFDAFNN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4MHR3EQTPWOTF5RB3ZJXFDAFNN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4MHR3EQTPWOTF5RB3ZJXFDAFNN/action/storage_attestation","attest_author":"https://pith.science/pith/4MHR3EQTPWOTF5RB3ZJXFDAFNN/action/author_attestation","sign_citation":"https://pith.science/pith/4MHR3EQTPWOTF5RB3ZJXFDAFNN/action/citation_signature","submit_replication":"https://pith.science/pith/4MHR3EQTPWOTF5RB3ZJXFDAFNN/action/replication_record"}},"created_at":"2026-07-05T09:34:47.755551+00:00","updated_at":"2026-07-05T09:34:47.755551+00:00"}