{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:3HBLRQPHZ5PT7QT7J6MNCD2PBT","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"184a8c5bf20a96648214c8b1ecf0e441da3b0bac29299f046e9f659c5b182393","cross_cats_sorted":["cs.AI","cs.DC"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-05T03:47:57Z","title_canon_sha256":"1e8a35ce44d5773751f8f000baf3afedff78c46e9b6019a392766d6e77929f0b"},"schema_version":"1.0","source":{"id":"2310.03294","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.03294","created_at":"2026-07-05T08:02:46Z"},{"alias_kind":"arxiv_version","alias_value":"2310.03294v2","created_at":"2026-07-05T08:02:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.03294","created_at":"2026-07-05T08:02:46Z"},{"alias_kind":"pith_short_12","alias_value":"3HBLRQPHZ5PT","created_at":"2026-07-05T08:02:46Z"},{"alias_kind":"pith_short_16","alias_value":"3HBLRQPHZ5PT7QT7","created_at":"2026-07-05T08:02:46Z"},{"alias_kind":"pith_short_8","alias_value":"3HBLRQPH","created_at":"2026-07-05T08:02:46Z"}],"graph_snapshots":[{"event_id":"sha256:04bb8bbb97f58c0e8ee70966c203fe115dbe2b34b0608cba8b179b71f2225d16","target":"graph","created_at":"2026-07-05T08:02:46Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2310.03294/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"FlashAttention (Dao, 2023) effectively reduces the quadratic peak memory usage to linear in training transformer-based large language models (LLMs) on a single GPU. In this paper, we introduce DISTFLASHATTN, a distributed memory-efficient attention mechanism optimized for long-context LLMs training. We propose three key techniques: token-level workload balancing, overlapping key-value communication, and a rematerialization-aware gradient checkpointing algorithm. We evaluate DISTFLASHATTN on Llama-7B and variants with sequence lengths from 32K to 512K. DISTFLASHATTN achieves 8x longer sequences","authors_text":"Anze Xie, Dacheng Li, Eric P. Xing, Hao Zhang, Ion Stoica, Joseph E. Gonzalez, Rulin Shao, Xuezhe Ma","cross_cats":["cs.AI","cs.DC"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-05T03:47:57Z","title":"DISTFLASHATTN: Distributed Memory-efficient Attention for Long-context LLMs Training"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.03294","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:5ff8c99075343a1cc2a93ad29fa6beb5c80969232046679d76e29307ae6e13bf","target":"record","created_at":"2026-07-05T08:02:46Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"184a8c5bf20a96648214c8b1ecf0e441da3b0bac29299f046e9f659c5b182393","cross_cats_sorted":["cs.AI","cs.DC"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-05T03:47:57Z","title_canon_sha256":"1e8a35ce44d5773751f8f000baf3afedff78c46e9b6019a392766d6e77929f0b"},"schema_version":"1.0","source":{"id":"2310.03294","kind":"arxiv","version":2}},"canonical_sha256":"d9c2b8c1e7cf5f3fc27f4f98d10f4f0cef963974c13153f0f3313dbe7442f984","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d9c2b8c1e7cf5f3fc27f4f98d10f4f0cef963974c13153f0f3313dbe7442f984","first_computed_at":"2026-07-05T08:02:46.702204Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:02:46.702204Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ezkHmf28zDH3ydHvzQHv3B7m5js9wFCYhg53RCTcrDRtZglzuHJb6k3nf1+wrQ5BsIxJCZakCIcI+X2CfSLlBA==","signature_status":"signed_v1","signed_at":"2026-07-05T08:02:46.702707Z","signed_message":"canonical_sha256_bytes"},"source_id":"2310.03294","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5ff8c99075343a1cc2a93ad29fa6beb5c80969232046679d76e29307ae6e13bf","sha256:04bb8bbb97f58c0e8ee70966c203fe115dbe2b34b0608cba8b179b71f2225d16"],"state_sha256":"c8e3ca3783083ec042972b20e38aa37e2e33142d103a4dade9cca16d4b6dec85"}