{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:HBYOTPW5VODT2VPHKJGV2DL5MI","short_pith_number":"pith:HBYOTPW5","canonical_record":{"source":{"id":"2507.19608","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-07-25T18:23:18Z","cross_cats_sorted":["eess.SP"],"title_canon_sha256":"84e196a8ea598b8c06903073505a9213cf46e93b579edc68845b559bde6778c4","abstract_canon_sha256":"ad9d2972f4e33ea4c08e72b8964bbebceb35c87fb7a32d5f1dd95f3da82b47d0"},"schema_version":"1.0"},"canonical_sha256":"3870e9beddab873d55e7524d5d0d7d6203370f8d44482880448937b08a83a682","source":{"kind":"arxiv","id":"2507.19608","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.19608","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"arxiv_version","alias_value":"2507.19608v1","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.19608","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"pith_short_12","alias_value":"HBYOTPW5VODT","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"pith_short_16","alias_value":"HBYOTPW5VODT2VPH","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"pith_short_8","alias_value":"HBYOTPW5","created_at":"2026-07-05T11:43:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:HBYOTPW5VODT2VPHKJGV2DL5MI","target":"record","payload":{"canonical_record":{"source":{"id":"2507.19608","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-07-25T18:23:18Z","cross_cats_sorted":["eess.SP"],"title_canon_sha256":"84e196a8ea598b8c06903073505a9213cf46e93b579edc68845b559bde6778c4","abstract_canon_sha256":"ad9d2972f4e33ea4c08e72b8964bbebceb35c87fb7a32d5f1dd95f3da82b47d0"},"schema_version":"1.0"},"canonical_sha256":"3870e9beddab873d55e7524d5d0d7d6203370f8d44482880448937b08a83a682","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:43:37.505239Z","signature_b64":"6YKn6p727wXH75a3BaWWMiQDX6blnczhRzXdCh/dQtmRhDSPlInlfoqW7YDgvrOLwFHyXgqMylQOxEQdghyZBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3870e9beddab873d55e7524d5d0d7d6203370f8d44482880448937b08a83a682","last_reissued_at":"2026-07-05T11:43:37.504724Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:43:37.504724Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2507.19608","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":"2026-07-05T11:43:37Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xzf/lIbMwnaRaiaMYIQpu8sNyylrpIc5m7jYE2wEwn+Lf4Zg9XImvH0t2e07jiGczrVpQ8fGvoyxvTioRNkDAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T18:39:17.129380Z"},"content_sha256":"ad18132ac947ea533f15288c1988b41470bd61892bcd4d9107fd3a1d8527d9e9","schema_version":"1.0","event_id":"sha256:ad18132ac947ea533f15288c1988b41470bd61892bcd4d9107fd3a1d8527d9e9"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:HBYOTPW5VODT2VPHKJGV2DL5MI","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"DeltaLLM: A Training-Free Framework Exploiting Temporal Sparsity for Efficient Edge LLM Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.SP"],"primary_cat":"cs.AI","authors_text":"Chang Gao, Jiawen Qi, Qinyu Chen, Zhaochun Ren","submitted_at":"2025-07-25T18:23:18Z","abstract_excerpt":"Deploying Large Language Models (LLMs) on edge devices remains challenging due to their quadratically increasing computations with the sequence length. Existing studies for dynamic attention pruning are designed for hardware with massively parallel computation capabilities, such as GPUs or TPUs, and aim at long context lengths (e.g., 64K), making them unsuitable for edge scenarios. We present DeltaLLM, a training-free framework that exploits temporal sparsity in attention patterns to enable efficient LLM inference across both the prefilling and decoding stages, on resource-constrained edge dev"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.19608","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/2507.19608/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":"2026-07-05T11:43:37Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"qMjNylEvh/jdmT2JHygIpv3mR05v10vgsITvsEzBSQQsuqSM51XMIbYxyK5k3eLEBGvMBPLSOH6JdfHKvoVOAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T18:39:17.130598Z"},"content_sha256":"5f03c7fd1bfc9c5d892ee79ba365a511fbf3e741fcbfc4c3633ab0871dc0fa00","schema_version":"1.0","event_id":"sha256:5f03c7fd1bfc9c5d892ee79ba365a511fbf3e741fcbfc4c3633ab0871dc0fa00"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/HBYOTPW5VODT2VPHKJGV2DL5MI/bundle.json","state_url":"https://pith.science/pith/HBYOTPW5VODT2VPHKJGV2DL5MI/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/HBYOTPW5VODT2VPHKJGV2DL5MI/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":"2026-08-08T18:39:17Z","links":{"resolver":"https://pith.science/pith/HBYOTPW5VODT2VPHKJGV2DL5MI","bundle":"https://pith.science/pith/HBYOTPW5VODT2VPHKJGV2DL5MI/bundle.json","state":"https://pith.science/pith/HBYOTPW5VODT2VPHKJGV2DL5MI/state.json","well_known_bundle":"https://pith.science/.well-known/pith/HBYOTPW5VODT2VPHKJGV2DL5MI/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:HBYOTPW5VODT2VPHKJGV2DL5MI","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":"ad9d2972f4e33ea4c08e72b8964bbebceb35c87fb7a32d5f1dd95f3da82b47d0","cross_cats_sorted":["eess.SP"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-07-25T18:23:18Z","title_canon_sha256":"84e196a8ea598b8c06903073505a9213cf46e93b579edc68845b559bde6778c4"},"schema_version":"1.0","source":{"id":"2507.19608","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2507.19608","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"arxiv_version","alias_value":"2507.19608v1","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.19608","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"pith_short_12","alias_value":"HBYOTPW5VODT","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"pith_short_16","alias_value":"HBYOTPW5VODT2VPH","created_at":"2026-07-05T11:43:37Z"},{"alias_kind":"pith_short_8","alias_value":"HBYOTPW5","created_at":"2026-07-05T11:43:37Z"}],"graph_snapshots":[{"event_id":"sha256:5f03c7fd1bfc9c5d892ee79ba365a511fbf3e741fcbfc4c3633ab0871dc0fa00","target":"graph","created_at":"2026-07-05T11:43:37Z","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/2507.19608/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Deploying Large Language Models (LLMs) on edge devices remains challenging due to their quadratically increasing computations with the sequence length. Existing studies for dynamic attention pruning are designed for hardware with massively parallel computation capabilities, such as GPUs or TPUs, and aim at long context lengths (e.g., 64K), making them unsuitable for edge scenarios. We present DeltaLLM, a training-free framework that exploits temporal sparsity in attention patterns to enable efficient LLM inference across both the prefilling and decoding stages, on resource-constrained edge dev","authors_text":"Chang Gao, Jiawen Qi, Qinyu Chen, Zhaochun Ren","cross_cats":["eess.SP"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-07-25T18:23:18Z","title":"DeltaLLM: A Training-Free Framework Exploiting Temporal Sparsity for Efficient Edge LLM Inference"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.19608","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:ad18132ac947ea533f15288c1988b41470bd61892bcd4d9107fd3a1d8527d9e9","target":"record","created_at":"2026-07-05T11:43:37Z","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":"ad9d2972f4e33ea4c08e72b8964bbebceb35c87fb7a32d5f1dd95f3da82b47d0","cross_cats_sorted":["eess.SP"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-07-25T18:23:18Z","title_canon_sha256":"84e196a8ea598b8c06903073505a9213cf46e93b579edc68845b559bde6778c4"},"schema_version":"1.0","source":{"id":"2507.19608","kind":"arxiv","version":1}},"canonical_sha256":"3870e9beddab873d55e7524d5d0d7d6203370f8d44482880448937b08a83a682","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3870e9beddab873d55e7524d5d0d7d6203370f8d44482880448937b08a83a682","first_computed_at":"2026-07-05T11:43:37.504724Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:43:37.504724Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"6YKn6p727wXH75a3BaWWMiQDX6blnczhRzXdCh/dQtmRhDSPlInlfoqW7YDgvrOLwFHyXgqMylQOxEQdghyZBw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:43:37.505239Z","signed_message":"canonical_sha256_bytes"},"source_id":"2507.19608","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ad18132ac947ea533f15288c1988b41470bd61892bcd4d9107fd3a1d8527d9e9","sha256:5f03c7fd1bfc9c5d892ee79ba365a511fbf3e741fcbfc4c3633ab0871dc0fa00"],"state_sha256":"f897e24825e81377dfd16678dbf9120ce94a43b34c3fe5ab39b85481632a8b71"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"af1TD3/Cn5dWopXfSzCVYwjWM2ZE4y/g1QSag2e1BVdcDmRokGTg4XJt7YkJep2Qq8TMdZ7m2sRQ3ooCTThCDg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T18:39:17.136400Z","bundle_sha256":"30ff593d5766c453ff959757aa65df909ce65db423f9f6a87aecca4911716e8d"}}