{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:WJ72VQG4FSVLB6L4Y2VSUWYBBP","short_pith_number":"pith:WJ72VQG4","schema_version":"1.0","canonical_sha256":"b27faac0dc2caab0f97cc6ab2a5b010bfce60dbf5177deeea06b4b90e7c9b84d","source":{"kind":"arxiv","id":"2511.21513","version":2},"attestation_state":"computed","paper":{"title":"IntAttention: A Fully Integer Attention Pipeline for Efficient Edge Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Haibo Feng, Hanyang Peng, Shiqi Yu, Wanli Zhong, Zirui Zhou","submitted_at":"2025-11-26T15:46:22Z","abstract_excerpt":"Deploying Transformer models on edge devices is limited by latency and energy budgets. While INT8 quantization effectively accelerates the primary matrix multiplications, it exposes the softmax-related path as the dominant bottleneck. This stage incurs a costly dequantize -> softmax -> requantize detour, which can account for up to 65% of total attention latency and disrupts the end-to-end integer dataflow critical for edge hardware efficiency. To address this limitation, we present IntAttention, the first fully integer attention pipeline that serves as a training-free drop-in replacement. At "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2511.21513","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-11-26T15:46:22Z","cross_cats_sorted":[],"title_canon_sha256":"6f3b436bac0eedfc4a372f954b1785be402f8b79fce4a9ad18c6197d69299192","abstract_canon_sha256":"691a2c7ff190d200a23993a5b704a0599c7fbf1093f6cb6dc533e6c04581f9a3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-01T01:02:26.149460Z","signature_b64":"85kNp7pLiR67t8tOQfrKU5AfJ5siGE58m1CE0fQQDLQKG+AizF8b0u8F1znMmni07IeMhpoiEguz1f73jXAfDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b27faac0dc2caab0f97cc6ab2a5b010bfce60dbf5177deeea06b4b90e7c9b84d","last_reissued_at":"2026-06-01T01:02:26.148517Z","signature_status":"signed_v1","first_computed_at":"2026-06-01T01:02:26.148517Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"IntAttention: A Fully Integer Attention Pipeline for Efficient Edge Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Haibo Feng, Hanyang Peng, Shiqi Yu, Wanli Zhong, Zirui Zhou","submitted_at":"2025-11-26T15:46:22Z","abstract_excerpt":"Deploying Transformer models on edge devices is limited by latency and energy budgets. While INT8 quantization effectively accelerates the primary matrix multiplications, it exposes the softmax-related path as the dominant bottleneck. This stage incurs a costly dequantize -> softmax -> requantize detour, which can account for up to 65% of total attention latency and disrupts the end-to-end integer dataflow critical for edge hardware efficiency. To address this limitation, we present IntAttention, the first fully integer attention pipeline that serves as a training-free drop-in replacement. At "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.21513","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/2511.21513/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2511.21513","created_at":"2026-06-01T01:02:26.148651+00:00"},{"alias_kind":"arxiv_version","alias_value":"2511.21513v2","created_at":"2026-06-01T01:02:26.148651+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.21513","created_at":"2026-06-01T01:02:26.148651+00:00"},{"alias_kind":"pith_short_12","alias_value":"WJ72VQG4FSVL","created_at":"2026-06-01T01:02:26.148651+00:00"},{"alias_kind":"pith_short_16","alias_value":"WJ72VQG4FSVLB6L4","created_at":"2026-06-01T01:02:26.148651+00:00"},{"alias_kind":"pith_short_8","alias_value":"WJ72VQG4","created_at":"2026-06-01T01:02:26.148651+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WJ72VQG4FSVLB6L4Y2VSUWYBBP","json":"https://pith.science/pith/WJ72VQG4FSVLB6L4Y2VSUWYBBP.json","graph_json":"https://pith.science/api/pith-number/WJ72VQG4FSVLB6L4Y2VSUWYBBP/graph.json","events_json":"https://pith.science/api/pith-number/WJ72VQG4FSVLB6L4Y2VSUWYBBP/events.json","paper":"https://pith.science/paper/WJ72VQG4"},"agent_actions":{"view_html":"https://pith.science/pith/WJ72VQG4FSVLB6L4Y2VSUWYBBP","download_json":"https://pith.science/pith/WJ72VQG4FSVLB6L4Y2VSUWYBBP.json","view_paper":"https://pith.science/paper/WJ72VQG4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2511.21513&json=true","fetch_graph":"https://pith.science/api/pith-number/WJ72VQG4FSVLB6L4Y2VSUWYBBP/graph.json","fetch_events":"https://pith.science/api/pith-number/WJ72VQG4FSVLB6L4Y2VSUWYBBP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WJ72VQG4FSVLB6L4Y2VSUWYBBP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WJ72VQG4FSVLB6L4Y2VSUWYBBP/action/storage_attestation","attest_author":"https://pith.science/pith/WJ72VQG4FSVLB6L4Y2VSUWYBBP/action/author_attestation","sign_citation":"https://pith.science/pith/WJ72VQG4FSVLB6L4Y2VSUWYBBP/action/citation_signature","submit_replication":"https://pith.science/pith/WJ72VQG4FSVLB6L4Y2VSUWYBBP/action/replication_record"}},"created_at":"2026-06-01T01:02:26.148651+00:00","updated_at":"2026-06-01T01:02:26.148651+00:00"}