{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:65TGQBRZONQT7GG5OESB7FI7SD","short_pith_number":"pith:65TGQBRZ","schema_version":"1.0","canonical_sha256":"f76668063973613f98dd71241f951f90d344246c6c43a109eb9a4bb5472a4f13","source":{"kind":"arxiv","id":"2302.06646","version":1},"attestation_state":"computed","paper":{"title":"Simple Hardware-Efficient Long Convolutions for Sequence Modeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Armin W. Thomas, Atri Rudra, Christopher R\\'e, Daniel Y. Fu, Elliot L. Epstein, Eric Nguyen, Michael Zhang, Tri Dao","submitted_at":"2023-02-13T19:19:23Z","abstract_excerpt":"State space models (SSMs) have high performance on long sequence modeling but require sophisticated initialization techniques and specialized implementations for high quality and runtime performance. We study whether a simple alternative can match SSMs in performance and efficiency: directly learning long convolutions over the sequence. We find that a key requirement to achieving high performance is keeping the convolution kernels smooth. We find that simple interventions--such as squashing the kernel weights--result in smooth kernels and recover SSM performance on a range of tasks including t"},"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":"2302.06646","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-02-13T19:19:23Z","cross_cats_sorted":[],"title_canon_sha256":"133ab6398ed266797acf2ab87b4f2cc864e6c1d2ee2cbbed06e8aaaabaa94af9","abstract_canon_sha256":"02c86548b6e8f5c56e18dfc0dbebe9a93b62334b3a1434aa0b0b904e772b8230"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:41:42.335852Z","signature_b64":"d+oWivVqdGEamkLrEjosZhdeJ5GVUsoE+4uy/TdC9wYNXN0XQ1pseYb/Z786RCMJRINic+12O6+8NVblzL2gDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f76668063973613f98dd71241f951f90d344246c6c43a109eb9a4bb5472a4f13","last_reissued_at":"2026-07-05T05:41:42.335367Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:41:42.335367Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Simple Hardware-Efficient Long Convolutions for Sequence Modeling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Armin W. Thomas, Atri Rudra, Christopher R\\'e, Daniel Y. Fu, Elliot L. Epstein, Eric Nguyen, Michael Zhang, Tri Dao","submitted_at":"2023-02-13T19:19:23Z","abstract_excerpt":"State space models (SSMs) have high performance on long sequence modeling but require sophisticated initialization techniques and specialized implementations for high quality and runtime performance. We study whether a simple alternative can match SSMs in performance and efficiency: directly learning long convolutions over the sequence. We find that a key requirement to achieving high performance is keeping the convolution kernels smooth. We find that simple interventions--such as squashing the kernel weights--result in smooth kernels and recover SSM performance on a range of tasks including t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.06646","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/2302.06646/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":"2302.06646","created_at":"2026-07-05T05:41:42.335428+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.06646v1","created_at":"2026-07-05T05:41:42.335428+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.06646","created_at":"2026-07-05T05:41:42.335428+00:00"},{"alias_kind":"pith_short_12","alias_value":"65TGQBRZONQT","created_at":"2026-07-05T05:41:42.335428+00:00"},{"alias_kind":"pith_short_16","alias_value":"65TGQBRZONQT7GG5","created_at":"2026-07-05T05:41:42.335428+00:00"},{"alias_kind":"pith_short_8","alias_value":"65TGQBRZ","created_at":"2026-07-05T05:41:42.335428+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.11275","citing_title":"RoVE: Rotary Value Embeddings Attention for Relative Position-dependent Value Pathways","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21070","citing_title":"Towards Understanding Self-Pretraining for Sequence Classification","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2312.06635","citing_title":"Gated Linear Attention Transformers with Hardware-Efficient Training","ref_index":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/65TGQBRZONQT7GG5OESB7FI7SD","json":"https://pith.science/pith/65TGQBRZONQT7GG5OESB7FI7SD.json","graph_json":"https://pith.science/api/pith-number/65TGQBRZONQT7GG5OESB7FI7SD/graph.json","events_json":"https://pith.science/api/pith-number/65TGQBRZONQT7GG5OESB7FI7SD/events.json","paper":"https://pith.science/paper/65TGQBRZ"},"agent_actions":{"view_html":"https://pith.science/pith/65TGQBRZONQT7GG5OESB7FI7SD","download_json":"https://pith.science/pith/65TGQBRZONQT7GG5OESB7FI7SD.json","view_paper":"https://pith.science/paper/65TGQBRZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.06646&json=true","fetch_graph":"https://pith.science/api/pith-number/65TGQBRZONQT7GG5OESB7FI7SD/graph.json","fetch_events":"https://pith.science/api/pith-number/65TGQBRZONQT7GG5OESB7FI7SD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/65TGQBRZONQT7GG5OESB7FI7SD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/65TGQBRZONQT7GG5OESB7FI7SD/action/storage_attestation","attest_author":"https://pith.science/pith/65TGQBRZONQT7GG5OESB7FI7SD/action/author_attestation","sign_citation":"https://pith.science/pith/65TGQBRZONQT7GG5OESB7FI7SD/action/citation_signature","submit_replication":"https://pith.science/pith/65TGQBRZONQT7GG5OESB7FI7SD/action/replication_record"}},"created_at":"2026-07-05T05:41:42.335428+00:00","updated_at":"2026-07-05T05:41:42.335428+00:00"}