{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:SCX77GS2WHUHV6HOLSNIVMYFJQ","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":"7a8511ce26fe85f7478da28c4bc7f44c147f28b1cb6ec22e87dd996dbe325a80","cross_cats_sorted":["cs.PF"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.DC","submitted_at":"2024-03-18T02:30:23Z","title_canon_sha256":"13262b41a41ad360510fd188c9f25770a6bb1847e09b407decea549eeefd2f62"},"schema_version":"1.0","source":{"id":"2403.11421","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.11421","created_at":"2026-07-05T07:57:16Z"},{"alias_kind":"arxiv_version","alias_value":"2403.11421v1","created_at":"2026-07-05T07:57:16Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.11421","created_at":"2026-07-05T07:57:16Z"},{"alias_kind":"pith_short_12","alias_value":"SCX77GS2WHUH","created_at":"2026-07-05T07:57:16Z"},{"alias_kind":"pith_short_16","alias_value":"SCX77GS2WHUHV6HO","created_at":"2026-07-05T07:57:16Z"},{"alias_kind":"pith_short_8","alias_value":"SCX77GS2","created_at":"2026-07-05T07:57:16Z"}],"graph_snapshots":[{"event_id":"sha256:60fd648ecdc0f29feb0de273cf88c67efa3b07c05eeaabe397c4ad017550b03f","target":"graph","created_at":"2026-07-05T07:57:16Z","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/2403.11421/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Cost of serving large language models (LLM) is high, but the expensive and scarce GPUs are poorly efficient when generating tokens sequentially, unless the batch of sequences is enlarged. However, the batch size is limited by some constantly reused intermediate results, namely KV-Cache. They occupy too much memory to fit more sequences into a GPU simultaneously. While they could be offloaded to host memory, the CPU-GPU bandwidth is an inevitable bottleneck.\n  We find a way to decompose the transformer models into two parts of different characteristics, one of which includes the memory-bound KV","authors_text":"Jiaao He, Jidong Zhai","cross_cats":["cs.PF"],"headline":"","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.DC","submitted_at":"2024-03-18T02:30:23Z","title":"FastDecode: High-Throughput GPU-Efficient LLM Serving using Heterogeneous Pipelines"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.11421","kind":"arxiv","version":1},"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:beed1ab1b2d18de6c525089606504c386048e511fe027338da1362efa48fd1a7","target":"record","created_at":"2026-07-05T07:57:16Z","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":"7a8511ce26fe85f7478da28c4bc7f44c147f28b1cb6ec22e87dd996dbe325a80","cross_cats_sorted":["cs.PF"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.DC","submitted_at":"2024-03-18T02:30:23Z","title_canon_sha256":"13262b41a41ad360510fd188c9f25770a6bb1847e09b407decea549eeefd2f62"},"schema_version":"1.0","source":{"id":"2403.11421","kind":"arxiv","version":1}},"canonical_sha256":"90afff9a5ab1e87af8ee5c9a8ab3054c0ca20557e23488701e4dd5b4484abef4","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"90afff9a5ab1e87af8ee5c9a8ab3054c0ca20557e23488701e4dd5b4484abef4","first_computed_at":"2026-07-05T07:57:16.179692Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:57:16.179692Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Ia6wYglDaDuZyN7blhoQBlCSWMl+Wdf8FD2umNf1hjlppg9ljLB6ThYaZiUYcMeobxG41/tZKfWCx0A1DCShBw==","signature_status":"signed_v1","signed_at":"2026-07-05T07:57:16.180116Z","signed_message":"canonical_sha256_bytes"},"source_id":"2403.11421","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:beed1ab1b2d18de6c525089606504c386048e511fe027338da1362efa48fd1a7","sha256:60fd648ecdc0f29feb0de273cf88c67efa3b07c05eeaabe397c4ad017550b03f"],"state_sha256":"59035508f5cda8ef4908182306571424302e3d843c3a8089c4109f7c3475df19"}