{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:ELGGVISF33OF6S3KHLH54JN4TO","short_pith_number":"pith:ELGGVISF","schema_version":"1.0","canonical_sha256":"22cc6aa245dedc5f4b6a3acfde25bc9b95f449814fccfe15cc81c3eae32487c0","source":{"kind":"arxiv","id":"2009.08605","version":1},"attestation_state":"computed","paper":{"title":"Hardware Accelerator for Multi-Head Attention and Position-Wise Feed-Forward in the Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AR"],"primary_cat":"eess.SP","authors_text":"Jun Lin, Meiqi Wang, Shuang Liang, Siyuan Lu, Zhongfeng Wang","submitted_at":"2020-09-18T03:13:19Z","abstract_excerpt":"Designing hardware accelerators for deep neural networks (DNNs) has been much desired. Nonetheless, most of these existing accelerators are built for either convolutional neural networks (CNNs) or recurrent neural networks (RNNs). Recently, the Transformer model is replacing the RNN in the natural language processing (NLP) area. However, because of intensive matrix computations and complicated data flow being involved, the hardware design for the Transformer model has never been reported. In this paper, we propose the first hardware accelerator for two key components, i.e., the multi-head atte"},"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":"2009.08605","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"eess.SP","submitted_at":"2020-09-18T03:13:19Z","cross_cats_sorted":["cs.AR"],"title_canon_sha256":"423a611abab646092e2e19b15ff76b111f93726dda340ef556ca2ee16664ebf4","abstract_canon_sha256":"2029c3237a930472f55caf245943b4c0be1688fc965f708678b637e4d8386730"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:36:20.165637Z","signature_b64":"JR1p7GomelqCgg7zZgjKb/xV6K1NEvUVbBFpTB0c4vxeswKQskBV/AEOn3AIe0OXW3Ti7UHbBCzSw1TmpGt6Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"22cc6aa245dedc5f4b6a3acfde25bc9b95f449814fccfe15cc81c3eae32487c0","last_reissued_at":"2026-07-05T01:36:20.165248Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:36:20.165248Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Hardware Accelerator for Multi-Head Attention and Position-Wise Feed-Forward in the Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AR"],"primary_cat":"eess.SP","authors_text":"Jun Lin, Meiqi Wang, Shuang Liang, Siyuan Lu, Zhongfeng Wang","submitted_at":"2020-09-18T03:13:19Z","abstract_excerpt":"Designing hardware accelerators for deep neural networks (DNNs) has been much desired. Nonetheless, most of these existing accelerators are built for either convolutional neural networks (CNNs) or recurrent neural networks (RNNs). Recently, the Transformer model is replacing the RNN in the natural language processing (NLP) area. However, because of intensive matrix computations and complicated data flow being involved, the hardware design for the Transformer model has never been reported. In this paper, we propose the first hardware accelerator for two key components, i.e., the multi-head atte"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2009.08605","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/2009.08605/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":"2009.08605","created_at":"2026-07-05T01:36:20.165301+00:00"},{"alias_kind":"arxiv_version","alias_value":"2009.08605v1","created_at":"2026-07-05T01:36:20.165301+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2009.08605","created_at":"2026-07-05T01:36:20.165301+00:00"},{"alias_kind":"pith_short_12","alias_value":"ELGGVISF33OF","created_at":"2026-07-05T01:36:20.165301+00:00"},{"alias_kind":"pith_short_16","alias_value":"ELGGVISF33OF6S3K","created_at":"2026-07-05T01:36:20.165301+00:00"},{"alias_kind":"pith_short_8","alias_value":"ELGGVISF","created_at":"2026-07-05T01:36:20.165301+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19913","citing_title":"Design and Evaluation of Energy-Efficient Whisper Dot-Product Kernel Offloading on a CGLA Architecture","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ELGGVISF33OF6S3KHLH54JN4TO","json":"https://pith.science/pith/ELGGVISF33OF6S3KHLH54JN4TO.json","graph_json":"https://pith.science/api/pith-number/ELGGVISF33OF6S3KHLH54JN4TO/graph.json","events_json":"https://pith.science/api/pith-number/ELGGVISF33OF6S3KHLH54JN4TO/events.json","paper":"https://pith.science/paper/ELGGVISF"},"agent_actions":{"view_html":"https://pith.science/pith/ELGGVISF33OF6S3KHLH54JN4TO","download_json":"https://pith.science/pith/ELGGVISF33OF6S3KHLH54JN4TO.json","view_paper":"https://pith.science/paper/ELGGVISF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2009.08605&json=true","fetch_graph":"https://pith.science/api/pith-number/ELGGVISF33OF6S3KHLH54JN4TO/graph.json","fetch_events":"https://pith.science/api/pith-number/ELGGVISF33OF6S3KHLH54JN4TO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ELGGVISF33OF6S3KHLH54JN4TO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ELGGVISF33OF6S3KHLH54JN4TO/action/storage_attestation","attest_author":"https://pith.science/pith/ELGGVISF33OF6S3KHLH54JN4TO/action/author_attestation","sign_citation":"https://pith.science/pith/ELGGVISF33OF6S3KHLH54JN4TO/action/citation_signature","submit_replication":"https://pith.science/pith/ELGGVISF33OF6S3KHLH54JN4TO/action/replication_record"}},"created_at":"2026-07-05T01:36:20.165301+00:00","updated_at":"2026-07-05T01:36:20.165301+00:00"}