{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:XVWDAH5GC36DTS2T7J6W5DSCAV","short_pith_number":"pith:XVWDAH5G","canonical_record":{"source":{"id":"2503.18292","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-03-24T02:28:04Z","cross_cats_sorted":[],"title_canon_sha256":"444c356835798778140757189ea0836fd864315d40e0500bbf235892ce8bf86b","abstract_canon_sha256":"74762168eb32975437d8d7e79aa135b9285314f366716f1912a4b814e0ef4df5"},"schema_version":"1.0"},"canonical_sha256":"bd6c301fa616fc39cb53fa7d6e8e420560a8485ced08c0c5e74eb09a1f226f63","source":{"kind":"arxiv","id":"2503.18292","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.18292","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"arxiv_version","alias_value":"2503.18292v1","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.18292","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"pith_short_12","alias_value":"XVWDAH5GC36D","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"pith_short_16","alias_value":"XVWDAH5GC36DTS2T","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"pith_short_8","alias_value":"XVWDAH5G","created_at":"2026-07-05T10:38:07Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:XVWDAH5GC36DTS2T7J6W5DSCAV","target":"record","payload":{"canonical_record":{"source":{"id":"2503.18292","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-03-24T02:28:04Z","cross_cats_sorted":[],"title_canon_sha256":"444c356835798778140757189ea0836fd864315d40e0500bbf235892ce8bf86b","abstract_canon_sha256":"74762168eb32975437d8d7e79aa135b9285314f366716f1912a4b814e0ef4df5"},"schema_version":"1.0"},"canonical_sha256":"bd6c301fa616fc39cb53fa7d6e8e420560a8485ced08c0c5e74eb09a1f226f63","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:38:07.263793Z","signature_b64":"Z3V4gXNYBGY6A3O8XBayzs3HzlyYil2kK1zUnZFI1D4DxzjQPgfsvKk9G2eVoomk4ByeY/ZWvyJhreESnZfgDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bd6c301fa616fc39cb53fa7d6e8e420560a8485ced08c0c5e74eb09a1f226f63","last_reissued_at":"2026-07-05T10:38:07.262942Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:38:07.262942Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2503.18292","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:38:07Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"RXiPJLUceuBbj9knnWXOA0UozcGhFf3U2E1Nsro7UlcTt8FgSGi2gWrMywHvVarijchz7oKSen/jPcywUVDVBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T09:36:34.150879Z"},"content_sha256":"04ae2f93dbaf1ef7ae19fbea5a377017bd9a4df93a854977e7c25c7d0cfa9900","schema_version":"1.0","event_id":"sha256:04ae2f93dbaf1ef7ae19fbea5a377017bd9a4df93a854977e7c25c7d0cfa9900"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:XVWDAH5GC36DTS2T7J6W5DSCAV","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Jenga: Effective Memory Management for Serving LLM with Heterogeneity","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Chen Zhang, Ion Stoica, Jidong Zhai, Joseph Gonzalez, Kaichao You, Kuntai Du, Mingsheng Long, Shu Liu, Woosuk Kwon, Xiangxi Mo, Xiaoxuan Liu, Yufeng Wang, Zhuohan Li","submitted_at":"2025-03-24T02:28:04Z","abstract_excerpt":"Large language models (LLMs) are widely used but expensive to run, especially as inference workloads grow. To lower costs, maximizing the request batch size by managing GPU memory efficiently is crucial. While PagedAttention has recently been proposed to improve the efficiency of memory management, we find that the growing heterogeneity in the embeddings dimensions, attention, and access patterns of modern LLM architectures introduces new challenges for memory allocation.\n  In this paper, we present Jenga, a novel memory allocation framework for heterogeneous embeddings in LLMs. Jenga tackles "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.18292","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/2503.18292/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:38:07Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"yPKeB3/KOtNcarfT0HYicL7JLNM5WMaonc91jNqHv6toX5LeoD2BypSdiPWo2QxbgAFqZWzFUtzbaQQiguvHCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T09:36:34.151565Z"},"content_sha256":"17d15578f1bc69f273ef626a44944a73e2a3519ac52b1a7c82a8cce43ca6955b","schema_version":"1.0","event_id":"sha256:17d15578f1bc69f273ef626a44944a73e2a3519ac52b1a7c82a8cce43ca6955b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XVWDAH5GC36DTS2T7J6W5DSCAV/bundle.json","state_url":"https://pith.science/pith/XVWDAH5GC36DTS2T7J6W5DSCAV/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XVWDAH5GC36DTS2T7J6W5DSCAV/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-04T09:36:34Z","links":{"resolver":"https://pith.science/pith/XVWDAH5GC36DTS2T7J6W5DSCAV","bundle":"https://pith.science/pith/XVWDAH5GC36DTS2T7J6W5DSCAV/bundle.json","state":"https://pith.science/pith/XVWDAH5GC36DTS2T7J6W5DSCAV/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XVWDAH5GC36DTS2T7J6W5DSCAV/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:XVWDAH5GC36DTS2T7J6W5DSCAV","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":"74762168eb32975437d8d7e79aa135b9285314f366716f1912a4b814e0ef4df5","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-03-24T02:28:04Z","title_canon_sha256":"444c356835798778140757189ea0836fd864315d40e0500bbf235892ce8bf86b"},"schema_version":"1.0","source":{"id":"2503.18292","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.18292","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"arxiv_version","alias_value":"2503.18292v1","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.18292","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"pith_short_12","alias_value":"XVWDAH5GC36D","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"pith_short_16","alias_value":"XVWDAH5GC36DTS2T","created_at":"2026-07-05T10:38:07Z"},{"alias_kind":"pith_short_8","alias_value":"XVWDAH5G","created_at":"2026-07-05T10:38:07Z"}],"graph_snapshots":[{"event_id":"sha256:17d15578f1bc69f273ef626a44944a73e2a3519ac52b1a7c82a8cce43ca6955b","target":"graph","created_at":"2026-07-05T10:38:07Z","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/2503.18292/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) are widely used but expensive to run, especially as inference workloads grow. To lower costs, maximizing the request batch size by managing GPU memory efficiently is crucial. While PagedAttention has recently been proposed to improve the efficiency of memory management, we find that the growing heterogeneity in the embeddings dimensions, attention, and access patterns of modern LLM architectures introduces new challenges for memory allocation.\n  In this paper, we present Jenga, a novel memory allocation framework for heterogeneous embeddings in LLMs. Jenga tackles ","authors_text":"Chen Zhang, Ion Stoica, Jidong Zhai, Joseph Gonzalez, Kaichao You, Kuntai Du, Mingsheng Long, Shu Liu, Woosuk Kwon, Xiangxi Mo, Xiaoxuan Liu, Yufeng Wang, Zhuohan Li","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-03-24T02:28:04Z","title":"Jenga: Effective Memory Management for Serving LLM with Heterogeneity"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.18292","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":"sha256:04ae2f93dbaf1ef7ae19fbea5a377017bd9a4df93a854977e7c25c7d0cfa9900","target":"record","created_at":"2026-07-05T10:38:07Z","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":"74762168eb32975437d8d7e79aa135b9285314f366716f1912a4b814e0ef4df5","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-03-24T02:28:04Z","title_canon_sha256":"444c356835798778140757189ea0836fd864315d40e0500bbf235892ce8bf86b"},"schema_version":"1.0","source":{"id":"2503.18292","kind":"arxiv","version":1}},"canonical_sha256":"bd6c301fa616fc39cb53fa7d6e8e420560a8485ced08c0c5e74eb09a1f226f63","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bd6c301fa616fc39cb53fa7d6e8e420560a8485ced08c0c5e74eb09a1f226f63","first_computed_at":"2026-07-05T10:38:07.262942Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:38:07.262942Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Z3V4gXNYBGY6A3O8XBayzs3HzlyYil2kK1zUnZFI1D4DxzjQPgfsvKk9G2eVoomk4ByeY/ZWvyJhreESnZfgDQ==","signature_status":"signed_v1","signed_at":"2026-07-05T10:38:07.263793Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.18292","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:04ae2f93dbaf1ef7ae19fbea5a377017bd9a4df93a854977e7c25c7d0cfa9900","sha256:17d15578f1bc69f273ef626a44944a73e2a3519ac52b1a7c82a8cce43ca6955b"],"state_sha256":"2da129870b6ca764f500aae2b79d04a6981f8b047b17809d6eaebda3484f984e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"cgnU+S2TJ1gDggl0ilvZz7/LHA8lmULCih0u01qTtwAtu//H3pNMP9hzXHYgT+QySlODf2wSeo7qjoeMo8hyBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-04T09:36:34.158129Z","bundle_sha256":"ef85010020aea281e7a93be0174d4928f60f34bb5f7fc978e8257f1e67eac0f7"}}