{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:R7BWDAPIIZQX2JQXDHHRLZJ5Q3","short_pith_number":"pith:R7BWDAPI","schema_version":"1.0","canonical_sha256":"8fc36181e846617d261719cf15e53d86d80955564c8d1a3479afb3ccfb02acac","source":{"kind":"arxiv","id":"2310.01082","version":2},"attestation_state":"computed","paper":{"title":"Linear attention is (maybe) all you need (to understand transformer optimization)","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","math.OC"],"primary_cat":"cs.LG","authors_text":"Ali Jadbabaie, Chulhee Yun, Kwangjun Ahn, Minhak Song, Suvrit Sra, Xiang Cheng","submitted_at":"2023-10-02T10:48:42Z","abstract_excerpt":"Transformer training is notoriously difficult, requiring a careful design of optimizers and use of various heuristics. We make progress towards understanding the subtleties of training Transformers by carefully studying a simple yet canonical linearized shallow Transformer model. Specifically, we train linear Transformers to solve regression tasks, inspired by J.~von Oswald et al.~(ICML 2023), and K.~Ahn et al.~(NeurIPS 2023). Most importantly, we observe that our proposed linearized models can reproduce several prominent aspects of Transformer training dynamics. Consequently, the results obta"},"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":"2310.01082","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-02T10:48:42Z","cross_cats_sorted":["cs.AI","math.OC"],"title_canon_sha256":"659128ca13bf2790e3429f3dc94867785bb22b7fde0a62c69bc03879a628364c","abstract_canon_sha256":"7227c8fef1f62770421543f9dd7dd94efe39574532ecbadd7e55b0a85a50b12b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:55:24.870715Z","signature_b64":"Xpk97+Ha9OJ11xV5qi408A78XEoDfsg7lvC5Ib4PxkcKgLpnmm4FAp0t9Yj9wxzw3tPi3PKz1fDfole9nr1GBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8fc36181e846617d261719cf15e53d86d80955564c8d1a3479afb3ccfb02acac","last_reissued_at":"2026-07-05T07:55:24.870278Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:55:24.870278Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Linear attention is (maybe) all you need (to understand transformer optimization)","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","math.OC"],"primary_cat":"cs.LG","authors_text":"Ali Jadbabaie, Chulhee Yun, Kwangjun Ahn, Minhak Song, Suvrit Sra, Xiang Cheng","submitted_at":"2023-10-02T10:48:42Z","abstract_excerpt":"Transformer training is notoriously difficult, requiring a careful design of optimizers and use of various heuristics. We make progress towards understanding the subtleties of training Transformers by carefully studying a simple yet canonical linearized shallow Transformer model. Specifically, we train linear Transformers to solve regression tasks, inspired by J.~von Oswald et al.~(ICML 2023), and K.~Ahn et al.~(NeurIPS 2023). Most importantly, we observe that our proposed linearized models can reproduce several prominent aspects of Transformer training dynamics. Consequently, the results obta"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.01082","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/2310.01082/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":"2310.01082","created_at":"2026-07-05T07:55:24.870343+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.01082v2","created_at":"2026-07-05T07:55:24.870343+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.01082","created_at":"2026-07-05T07:55:24.870343+00:00"},{"alias_kind":"pith_short_12","alias_value":"R7BWDAPIIZQX","created_at":"2026-07-05T07:55:24.870343+00:00"},{"alias_kind":"pith_short_16","alias_value":"R7BWDAPIIZQX2JQX","created_at":"2026-07-05T07:55:24.870343+00:00"},{"alias_kind":"pith_short_8","alias_value":"R7BWDAPI","created_at":"2026-07-05T07:55:24.870343+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.24513","citing_title":"Zeroth-Order Nonconvex Nonsmooth Optimization with Heavy-Tailed Noise","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11838","citing_title":"Gradient Clipping Beyond Vector Norms: A Spectral Approach for Matrix-Valued Parameters","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/R7BWDAPIIZQX2JQXDHHRLZJ5Q3","json":"https://pith.science/pith/R7BWDAPIIZQX2JQXDHHRLZJ5Q3.json","graph_json":"https://pith.science/api/pith-number/R7BWDAPIIZQX2JQXDHHRLZJ5Q3/graph.json","events_json":"https://pith.science/api/pith-number/R7BWDAPIIZQX2JQXDHHRLZJ5Q3/events.json","paper":"https://pith.science/paper/R7BWDAPI"},"agent_actions":{"view_html":"https://pith.science/pith/R7BWDAPIIZQX2JQXDHHRLZJ5Q3","download_json":"https://pith.science/pith/R7BWDAPIIZQX2JQXDHHRLZJ5Q3.json","view_paper":"https://pith.science/paper/R7BWDAPI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.01082&json=true","fetch_graph":"https://pith.science/api/pith-number/R7BWDAPIIZQX2JQXDHHRLZJ5Q3/graph.json","fetch_events":"https://pith.science/api/pith-number/R7BWDAPIIZQX2JQXDHHRLZJ5Q3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R7BWDAPIIZQX2JQXDHHRLZJ5Q3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R7BWDAPIIZQX2JQXDHHRLZJ5Q3/action/storage_attestation","attest_author":"https://pith.science/pith/R7BWDAPIIZQX2JQXDHHRLZJ5Q3/action/author_attestation","sign_citation":"https://pith.science/pith/R7BWDAPIIZQX2JQXDHHRLZJ5Q3/action/citation_signature","submit_replication":"https://pith.science/pith/R7BWDAPIIZQX2JQXDHHRLZJ5Q3/action/replication_record"}},"created_at":"2026-07-05T07:55:24.870343+00:00","updated_at":"2026-07-05T07:55:24.870343+00:00"}