{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FI7GUSWQVLHRF6V7CYU44CHS7Q","short_pith_number":"pith:FI7GUSWQ","schema_version":"1.0","canonical_sha256":"2a3e6a4ad0aacf12fabf1629ce08f2fc1ad2816c3b534b8112a20ac6cd48e1de","source":{"kind":"arxiv","id":"2602.22352","version":2},"attestation_state":"computed","paper":{"title":"GRAU: Generic Reconfigurable Activation Unit Design for Neural Network Hardware Accelerators","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.AR","authors_text":"Akash Kumar, Salim Ullah, Yuhao Liu","submitted_at":"2026-02-25T19:18:22Z","abstract_excerpt":"With the continuous growth of neural network scales, low-precision quantization is widely used in edge accelerators. Classic multi-threshold activation hardware requires 2^n thresholds for $n$-bit outputs, causing a rapid increase in hardware cost as precision increases. We propose a reconfigurable activation hardware, GRAU, based on piecewise linear fitting, where the segment slopes are approximated by powers of two. Our design requires only basic comparators and 1-bit right shifters, supporting mixed-precision quantization and nonlinear functions such as SiLU. Compared with multi-threshold a"},"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":"2602.22352","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AR","submitted_at":"2026-02-25T19:18:22Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"fa49e025f3c96d4f382278e20d1693261558e5a60801a47332dfdb076da0fea2","abstract_canon_sha256":"02dca1bc53e24726894dc3297aa1364a28420ead417a1544202b821e1cbc7a01"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-10T01:10:59.772364Z","signature_b64":"vW/oecCwCaHBR8bVS76FlNLGiosRgx6/NlR+8mBBDmIh7zP2h3u9MXuAzZQQlLc49wtfRJ4kq/NLkr8X711CDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2a3e6a4ad0aacf12fabf1629ce08f2fc1ad2816c3b534b8112a20ac6cd48e1de","last_reissued_at":"2026-06-10T01:10:59.771332Z","signature_status":"signed_v1","first_computed_at":"2026-06-10T01:10:59.771332Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GRAU: Generic Reconfigurable Activation Unit Design for Neural Network Hardware Accelerators","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.AR","authors_text":"Akash Kumar, Salim Ullah, Yuhao Liu","submitted_at":"2026-02-25T19:18:22Z","abstract_excerpt":"With the continuous growth of neural network scales, low-precision quantization is widely used in edge accelerators. Classic multi-threshold activation hardware requires 2^n thresholds for $n$-bit outputs, causing a rapid increase in hardware cost as precision increases. We propose a reconfigurable activation hardware, GRAU, based on piecewise linear fitting, where the segment slopes are approximated by powers of two. Our design requires only basic comparators and 1-bit right shifters, supporting mixed-precision quantization and nonlinear functions such as SiLU. Compared with multi-threshold a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.22352","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/2602.22352/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":"2602.22352","created_at":"2026-06-10T01:10:59.771527+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.22352v2","created_at":"2026-06-10T01:10:59.771527+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.22352","created_at":"2026-06-10T01:10:59.771527+00:00"},{"alias_kind":"pith_short_12","alias_value":"FI7GUSWQVLHR","created_at":"2026-06-10T01:10:59.771527+00:00"},{"alias_kind":"pith_short_16","alias_value":"FI7GUSWQVLHRF6V7","created_at":"2026-06-10T01:10:59.771527+00:00"},{"alias_kind":"pith_short_8","alias_value":"FI7GUSWQ","created_at":"2026-06-10T01:10:59.771527+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/FI7GUSWQVLHRF6V7CYU44CHS7Q","json":"https://pith.science/pith/FI7GUSWQVLHRF6V7CYU44CHS7Q.json","graph_json":"https://pith.science/api/pith-number/FI7GUSWQVLHRF6V7CYU44CHS7Q/graph.json","events_json":"https://pith.science/api/pith-number/FI7GUSWQVLHRF6V7CYU44CHS7Q/events.json","paper":"https://pith.science/paper/FI7GUSWQ"},"agent_actions":{"view_html":"https://pith.science/pith/FI7GUSWQVLHRF6V7CYU44CHS7Q","download_json":"https://pith.science/pith/FI7GUSWQVLHRF6V7CYU44CHS7Q.json","view_paper":"https://pith.science/paper/FI7GUSWQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.22352&json=true","fetch_graph":"https://pith.science/api/pith-number/FI7GUSWQVLHRF6V7CYU44CHS7Q/graph.json","fetch_events":"https://pith.science/api/pith-number/FI7GUSWQVLHRF6V7CYU44CHS7Q/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FI7GUSWQVLHRF6V7CYU44CHS7Q/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FI7GUSWQVLHRF6V7CYU44CHS7Q/action/storage_attestation","attest_author":"https://pith.science/pith/FI7GUSWQVLHRF6V7CYU44CHS7Q/action/author_attestation","sign_citation":"https://pith.science/pith/FI7GUSWQVLHRF6V7CYU44CHS7Q/action/citation_signature","submit_replication":"https://pith.science/pith/FI7GUSWQVLHRF6V7CYU44CHS7Q/action/replication_record"}},"created_at":"2026-06-10T01:10:59.771527+00:00","updated_at":"2026-06-10T01:10:59.771527+00:00"}