{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:B2FU44HFDPEAVWZWZRXQYFT27G","short_pith_number":"pith:B2FU44HF","schema_version":"1.0","canonical_sha256":"0e8b4e70e51bc80adb36cc6f0c167af996761d748bf2b628d1276cb9feb0c27c","source":{"kind":"arxiv","id":"2410.12707","version":1},"attestation_state":"computed","paper":{"title":"FusionLLM: A Decentralized LLM Training System on Geo-distributed GPUs with Adaptive Compression","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.DC","authors_text":"Amelie Chi Zhou, Bingsheng He, Bo Li, Kaiyong Zhao, Qiang Wang, Rongfei Zeng, Shaohuai Shi, Xiaowen Chu, Xinglin Pan, Xin He, Xueze Kang, Yiming Yin, Yuxin Wang, Zhenheng Tang","submitted_at":"2024-10-16T16:13:19Z","abstract_excerpt":"To alleviate hardware scarcity in training large deep neural networks (DNNs), particularly large language models (LLMs), we present FusionLLM, a decentralized training system designed and implemented for training DNNs using geo-distributed GPUs across different computing clusters or individual devices. Decentralized training faces significant challenges regarding system design and efficiency, including: 1) the need for remote automatic differentiation (RAD), 2) support for flexible model definitions and heterogeneous software, 3) heterogeneous hardware leading to low resource utilization or th"},"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":"2410.12707","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2024-10-16T16:13:19Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"be19a12a2e4679500a0ca9cf5a34de5853b32597afef4bd68ccb7b9f8bf5f1bb","abstract_canon_sha256":"ac9c1d8aff50c2e49078541315854e1936e237e97d8339d7064d457172f7bfbd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:21:32.690500Z","signature_b64":"7SdcjvVNUH+3qufvTvqC8mTsJwwTVc0uVN47J0gGZ8EIld+JTCqlMaJvqd1IhncWSSxF3kMQshjV79jhzkiQBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0e8b4e70e51bc80adb36cc6f0c167af996761d748bf2b628d1276cb9feb0c27c","last_reissued_at":"2026-07-05T09:21:32.690021Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:21:32.690021Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FusionLLM: A Decentralized LLM Training System on Geo-distributed GPUs with Adaptive Compression","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.DC","authors_text":"Amelie Chi Zhou, Bingsheng He, Bo Li, Kaiyong Zhao, Qiang Wang, Rongfei Zeng, Shaohuai Shi, Xiaowen Chu, Xinglin Pan, Xin He, Xueze Kang, Yiming Yin, Yuxin Wang, Zhenheng Tang","submitted_at":"2024-10-16T16:13:19Z","abstract_excerpt":"To alleviate hardware scarcity in training large deep neural networks (DNNs), particularly large language models (LLMs), we present FusionLLM, a decentralized training system designed and implemented for training DNNs using geo-distributed GPUs across different computing clusters or individual devices. Decentralized training faces significant challenges regarding system design and efficiency, including: 1) the need for remote automatic differentiation (RAD), 2) support for flexible model definitions and heterogeneous software, 3) heterogeneous hardware leading to low resource utilization or th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.12707","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/2410.12707/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":"2410.12707","created_at":"2026-07-05T09:21:32.690081+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.12707v1","created_at":"2026-07-05T09:21:32.690081+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.12707","created_at":"2026-07-05T09:21:32.690081+00:00"},{"alias_kind":"pith_short_12","alias_value":"B2FU44HFDPEA","created_at":"2026-07-05T09:21:32.690081+00:00"},{"alias_kind":"pith_short_16","alias_value":"B2FU44HFDPEAVWZW","created_at":"2026-07-05T09:21:32.690081+00:00"},{"alias_kind":"pith_short_8","alias_value":"B2FU44HF","created_at":"2026-07-05T09:21:32.690081+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24722","citing_title":"Decentralised AI Training and Inference with BlockTrain","ref_index":83,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24326","citing_title":"ScaleAcross Explorer: Exploring Communication Optimization for Scale-Across AI Model Training","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25375","citing_title":"Bandwidth-Aware and Cost-Efficient Pipeline Parallel Scheduling in Geo-Distributed LLM Training","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/B2FU44HFDPEAVWZWZRXQYFT27G","json":"https://pith.science/pith/B2FU44HFDPEAVWZWZRXQYFT27G.json","graph_json":"https://pith.science/api/pith-number/B2FU44HFDPEAVWZWZRXQYFT27G/graph.json","events_json":"https://pith.science/api/pith-number/B2FU44HFDPEAVWZWZRXQYFT27G/events.json","paper":"https://pith.science/paper/B2FU44HF"},"agent_actions":{"view_html":"https://pith.science/pith/B2FU44HFDPEAVWZWZRXQYFT27G","download_json":"https://pith.science/pith/B2FU44HFDPEAVWZWZRXQYFT27G.json","view_paper":"https://pith.science/paper/B2FU44HF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.12707&json=true","fetch_graph":"https://pith.science/api/pith-number/B2FU44HFDPEAVWZWZRXQYFT27G/graph.json","fetch_events":"https://pith.science/api/pith-number/B2FU44HFDPEAVWZWZRXQYFT27G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/B2FU44HFDPEAVWZWZRXQYFT27G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/B2FU44HFDPEAVWZWZRXQYFT27G/action/storage_attestation","attest_author":"https://pith.science/pith/B2FU44HFDPEAVWZWZRXQYFT27G/action/author_attestation","sign_citation":"https://pith.science/pith/B2FU44HFDPEAVWZWZRXQYFT27G/action/citation_signature","submit_replication":"https://pith.science/pith/B2FU44HFDPEAVWZWZRXQYFT27G/action/replication_record"}},"created_at":"2026-07-05T09:21:32.690081+00:00","updated_at":"2026-07-05T09:21:32.690081+00:00"}