{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:22TNDKQDRV7VYD7VVAUW4WJXOT","short_pith_number":"pith:22TNDKQD","canonical_record":{"source":{"id":"2501.08192","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-01-14T15:14:10Z","cross_cats_sorted":["cs.AR","cs.DC"],"title_canon_sha256":"853c71e6362e906f55e6f8647a1ac8ad315018fff81fe0ff93baf274b3ca1cf1","abstract_canon_sha256":"1b697fb0d6b66756b0aa6e1a7aae6ac04098ec6eb46932c1fee4dc02cd692d7b"},"schema_version":"1.0"},"canonical_sha256":"d6a6d1aa038d7f5c0ff5a8296e593774c6aa826c4d9a49e6cdd1ce84e91e5836","source":{"kind":"arxiv","id":"2501.08192","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.08192","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"arxiv_version","alias_value":"2501.08192v2","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.08192","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"pith_short_12","alias_value":"22TNDKQDRV7V","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"pith_short_16","alias_value":"22TNDKQDRV7VYD7V","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"pith_short_8","alias_value":"22TNDKQD","created_at":"2026-07-05T11:09:06Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:22TNDKQDRV7VYD7VVAUW4WJXOT","target":"record","payload":{"canonical_record":{"source":{"id":"2501.08192","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-01-14T15:14:10Z","cross_cats_sorted":["cs.AR","cs.DC"],"title_canon_sha256":"853c71e6362e906f55e6f8647a1ac8ad315018fff81fe0ff93baf274b3ca1cf1","abstract_canon_sha256":"1b697fb0d6b66756b0aa6e1a7aae6ac04098ec6eb46932c1fee4dc02cd692d7b"},"schema_version":"1.0"},"canonical_sha256":"d6a6d1aa038d7f5c0ff5a8296e593774c6aa826c4d9a49e6cdd1ce84e91e5836","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:09:06.522944Z","signature_b64":"WtJ7jSpb43xLrNObEj7wzqzcD/W0r2sljMJDQwvf6xL3VHwdviWDNkSNZ8ZbpgNeZYb9TSITMJrNFaRmmuV7Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d6a6d1aa038d7f5c0ff5a8296e593774c6aa826c4d9a49e6cdd1ce84e91e5836","last_reissued_at":"2026-07-05T11:09:06.522504Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:09:06.522504Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2501.08192","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:09:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5AXU2lt5KiIKxGUY9xaG++TeNafLvQ95xhTxWfOYfVDtwtFD42NiBf/VZj4TUixcZ/joHD145BQybjROvnyuBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-20T18:10:31.651171Z"},"content_sha256":"66240b7c7c1c5b99841d9b775cd917be1189bfde80af43618a2f131077a22bfd","schema_version":"1.0","event_id":"sha256:66240b7c7c1c5b99841d9b775cd917be1189bfde80af43618a2f131077a22bfd"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:22TNDKQDRV7VYD7VVAUW4WJXOT","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"PRESERVE: Prefetching Model Weights and KV-Cache in Distributed LLM Serving","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AR","cs.DC"],"primary_cat":"cs.AI","authors_text":"Ahmet Caner Y\\\"uz\\\"ug\\\"uler, Jiawei Zhuang, Lukas Cavigelli","submitted_at":"2025-01-14T15:14:10Z","abstract_excerpt":"Large language models (LLMs) are typically served from clusters of GPUs/NPUs that consist of large number of devices. Unfortunately, communication between these devices incurs significant overhead, increasing the inference latency and cost while limiting the scalability. Prior work addressed this issue by overlapping communication with compute, but has severe limitations due to the data dependencies between these operations. In this paper, we propose PRESERVE, a novel framework that prefetches model weights and KV-cache from off-chip HBM memory to the on-chip cache of AI accelerators during th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.08192","kind":"arxiv","version":2},"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/2501.08192/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":"2026-07-05T11:09:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"FFtLtnXWDM9uIYv+IrdFDtCvBayG9n+sD7aXqAwfPQbU8YNde1Nycur4vD+ney/e1CMGoQMlK+qmGN0GiHjsCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-20T18:10:31.651697Z"},"content_sha256":"24e4995813ad530a6ca99d574a4b334ce9eb919e6b3b463b12a48e622dcab0ba","schema_version":"1.0","event_id":"sha256:24e4995813ad530a6ca99d574a4b334ce9eb919e6b3b463b12a48e622dcab0ba"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/22TNDKQDRV7VYD7VVAUW4WJXOT/bundle.json","state_url":"https://pith.science/pith/22TNDKQDRV7VYD7VVAUW4WJXOT/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/22TNDKQDRV7VYD7VVAUW4WJXOT/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":"2026-08-20T18:10:31Z","links":{"resolver":"https://pith.science/pith/22TNDKQDRV7VYD7VVAUW4WJXOT","bundle":"https://pith.science/pith/22TNDKQDRV7VYD7VVAUW4WJXOT/bundle.json","state":"https://pith.science/pith/22TNDKQDRV7VYD7VVAUW4WJXOT/state.json","well_known_bundle":"https://pith.science/.well-known/pith/22TNDKQDRV7VYD7VVAUW4WJXOT/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:22TNDKQDRV7VYD7VVAUW4WJXOT","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":"1b697fb0d6b66756b0aa6e1a7aae6ac04098ec6eb46932c1fee4dc02cd692d7b","cross_cats_sorted":["cs.AR","cs.DC"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-01-14T15:14:10Z","title_canon_sha256":"853c71e6362e906f55e6f8647a1ac8ad315018fff81fe0ff93baf274b3ca1cf1"},"schema_version":"1.0","source":{"id":"2501.08192","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.08192","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"arxiv_version","alias_value":"2501.08192v2","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.08192","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"pith_short_12","alias_value":"22TNDKQDRV7V","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"pith_short_16","alias_value":"22TNDKQDRV7VYD7V","created_at":"2026-07-05T11:09:06Z"},{"alias_kind":"pith_short_8","alias_value":"22TNDKQD","created_at":"2026-07-05T11:09:06Z"}],"graph_snapshots":[{"event_id":"sha256:24e4995813ad530a6ca99d574a4b334ce9eb919e6b3b463b12a48e622dcab0ba","target":"graph","created_at":"2026-07-05T11:09:06Z","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/2501.08192/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) are typically served from clusters of GPUs/NPUs that consist of large number of devices. Unfortunately, communication between these devices incurs significant overhead, increasing the inference latency and cost while limiting the scalability. Prior work addressed this issue by overlapping communication with compute, but has severe limitations due to the data dependencies between these operations. In this paper, we propose PRESERVE, a novel framework that prefetches model weights and KV-cache from off-chip HBM memory to the on-chip cache of AI accelerators during th","authors_text":"Ahmet Caner Y\\\"uz\\\"ug\\\"uler, Jiawei Zhuang, Lukas Cavigelli","cross_cats":["cs.AR","cs.DC"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-01-14T15:14:10Z","title":"PRESERVE: Prefetching Model Weights and KV-Cache in Distributed LLM Serving"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.08192","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":"sha256:66240b7c7c1c5b99841d9b775cd917be1189bfde80af43618a2f131077a22bfd","target":"record","created_at":"2026-07-05T11:09:06Z","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":"1b697fb0d6b66756b0aa6e1a7aae6ac04098ec6eb46932c1fee4dc02cd692d7b","cross_cats_sorted":["cs.AR","cs.DC"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-01-14T15:14:10Z","title_canon_sha256":"853c71e6362e906f55e6f8647a1ac8ad315018fff81fe0ff93baf274b3ca1cf1"},"schema_version":"1.0","source":{"id":"2501.08192","kind":"arxiv","version":2}},"canonical_sha256":"d6a6d1aa038d7f5c0ff5a8296e593774c6aa826c4d9a49e6cdd1ce84e91e5836","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d6a6d1aa038d7f5c0ff5a8296e593774c6aa826c4d9a49e6cdd1ce84e91e5836","first_computed_at":"2026-07-05T11:09:06.522504Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:09:06.522504Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"WtJ7jSpb43xLrNObEj7wzqzcD/W0r2sljMJDQwvf6xL3VHwdviWDNkSNZ8ZbpgNeZYb9TSITMJrNFaRmmuV7Dw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:09:06.522944Z","signed_message":"canonical_sha256_bytes"},"source_id":"2501.08192","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:66240b7c7c1c5b99841d9b775cd917be1189bfde80af43618a2f131077a22bfd","sha256:24e4995813ad530a6ca99d574a4b334ce9eb919e6b3b463b12a48e622dcab0ba"],"state_sha256":"4ff518afae7fd0947bd41d1175206469cadbd64ba25567e371c032d815a86659"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"znNqfj5/JmVtrt/UFwaNc6uQNBb1m5iM+jpNTVPpRONw8xzVdu/bAIE4VYZGnMxRCKCU0+LpTvLBDM3gYEcjBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-20T18:10:31.657703Z","bundle_sha256":"1ff2a76cc1baefc3e79d2f23a9e61baf6cb1f35b28789723a505ca3f9be642d8"}}