{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:7CLUGEUBBYLZU4BHTSJD6B5V7T","short_pith_number":"pith:7CLUGEUB","schema_version":"1.0","canonical_sha256":"f8974312810e179a70279c923f07b5fcec245706796b8bd3d96d3e9ab806a7fc","source":{"kind":"arxiv","id":"2005.14187","version":1},"attestation_state":"computed","paper":{"title":"HAT: Hardware-Aware Transformers for Efficient Natural Language Processing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.NE"],"primary_cat":"cs.CL","authors_text":"Chuang Gan, Han Cai, Hanrui Wang, Ligeng Zhu, Song Han, Zhanghao Wu, Zhijian Liu","submitted_at":"2020-05-28T17:58:56Z","abstract_excerpt":"Transformers are ubiquitous in Natural Language Processing (NLP) tasks, but they are difficult to be deployed on hardware due to the intensive computation. To enable low-latency inference on resource-constrained hardware platforms, we propose to design Hardware-Aware Transformers (HAT) with neural architecture search. We first construct a large design space with $\\textit{arbitrary encoder-decoder attention}$ and $\\textit{heterogeneous layers}$. Then we train a $\\textit{SuperTransformer}$ that covers all candidates in the design space, and efficiently produces many $\\textit{SubTransformers}$ wi"},"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":"2005.14187","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-05-28T17:58:56Z","cross_cats_sorted":["cs.LG","cs.NE"],"title_canon_sha256":"939c040e3e689b324c5084bfde2a218a166bf4508499406ec3e7b22e75d71997","abstract_canon_sha256":"3258afe7634b1f573179af6ad7241913d5325e1afbea7ceb14ae54b87240382d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:04:09.792417Z","signature_b64":"PLW8Ij+eoSY+mGYXBsnzu34SIBJWSi+43YyP5HHsXgEoaTKUvQ669r8QUwI5qMe0CNoAlKBhgRDiU5NiGpjHBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f8974312810e179a70279c923f07b5fcec245706796b8bd3d96d3e9ab806a7fc","last_reissued_at":"2026-07-05T08:04:09.791932Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:04:09.791932Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HAT: Hardware-Aware Transformers for Efficient Natural Language Processing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.NE"],"primary_cat":"cs.CL","authors_text":"Chuang Gan, Han Cai, Hanrui Wang, Ligeng Zhu, Song Han, Zhanghao Wu, Zhijian Liu","submitted_at":"2020-05-28T17:58:56Z","abstract_excerpt":"Transformers are ubiquitous in Natural Language Processing (NLP) tasks, but they are difficult to be deployed on hardware due to the intensive computation. To enable low-latency inference on resource-constrained hardware platforms, we propose to design Hardware-Aware Transformers (HAT) with neural architecture search. We first construct a large design space with $\\textit{arbitrary encoder-decoder attention}$ and $\\textit{heterogeneous layers}$. Then we train a $\\textit{SuperTransformer}$ that covers all candidates in the design space, and efficiently produces many $\\textit{SubTransformers}$ wi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2005.14187","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/2005.14187/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":"2005.14187","created_at":"2026-07-05T08:04:09.791986+00:00"},{"alias_kind":"arxiv_version","alias_value":"2005.14187v1","created_at":"2026-07-05T08:04:09.791986+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2005.14187","created_at":"2026-07-05T08:04:09.791986+00:00"},{"alias_kind":"pith_short_12","alias_value":"7CLUGEUBBYLZ","created_at":"2026-07-05T08:04:09.791986+00:00"},{"alias_kind":"pith_short_16","alias_value":"7CLUGEUBBYLZU4BH","created_at":"2026-07-05T08:04:09.791986+00:00"},{"alias_kind":"pith_short_8","alias_value":"7CLUGEUB","created_at":"2026-07-05T08:04:09.791986+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.17653","citing_title":"LLMForge: Multi-Backend Hardware-Aware Neural Architecture Search with Infinite-Head Attention for Edge Language Models","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2510.14235","citing_title":"Spiking Neural Network Architecture Search: A Survey","ref_index":133,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7CLUGEUBBYLZU4BHTSJD6B5V7T","json":"https://pith.science/pith/7CLUGEUBBYLZU4BHTSJD6B5V7T.json","graph_json":"https://pith.science/api/pith-number/7CLUGEUBBYLZU4BHTSJD6B5V7T/graph.json","events_json":"https://pith.science/api/pith-number/7CLUGEUBBYLZU4BHTSJD6B5V7T/events.json","paper":"https://pith.science/paper/7CLUGEUB"},"agent_actions":{"view_html":"https://pith.science/pith/7CLUGEUBBYLZU4BHTSJD6B5V7T","download_json":"https://pith.science/pith/7CLUGEUBBYLZU4BHTSJD6B5V7T.json","view_paper":"https://pith.science/paper/7CLUGEUB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2005.14187&json=true","fetch_graph":"https://pith.science/api/pith-number/7CLUGEUBBYLZU4BHTSJD6B5V7T/graph.json","fetch_events":"https://pith.science/api/pith-number/7CLUGEUBBYLZU4BHTSJD6B5V7T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7CLUGEUBBYLZU4BHTSJD6B5V7T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7CLUGEUBBYLZU4BHTSJD6B5V7T/action/storage_attestation","attest_author":"https://pith.science/pith/7CLUGEUBBYLZU4BHTSJD6B5V7T/action/author_attestation","sign_citation":"https://pith.science/pith/7CLUGEUBBYLZU4BHTSJD6B5V7T/action/citation_signature","submit_replication":"https://pith.science/pith/7CLUGEUBBYLZU4BHTSJD6B5V7T/action/replication_record"}},"created_at":"2026-07-05T08:04:09.791986+00:00","updated_at":"2026-07-05T08:04:09.791986+00:00"}