{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:MLRPM4YIFRVCES4D2ZVBWTPLCG","short_pith_number":"pith:MLRPM4YI","schema_version":"1.0","canonical_sha256":"62e2f673082c6a224b83d66a1b4deb119b86423808ec51625f95f01db78f7c43","source":{"kind":"arxiv","id":"2308.16898","version":3},"attestation_state":"computed","paper":{"title":"Transformers as Support Vector Machines","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","math.OC"],"primary_cat":"cs.LG","authors_text":"Christos Thrampoulidis, Davoud Ataee Tarzanagh, Samet Oymak, Yingcong Li","submitted_at":"2023-08-31T17:57:50Z","abstract_excerpt":"Since its inception in \"Attention Is All You Need\", transformer architecture has led to revolutionary advancements in NLP. The attention layer within the transformer admits a sequence of input tokens $X$ and makes them interact through pairwise similarities computed as softmax$(XQK^\\top X^\\top)$, where $(K,Q)$ are the trainable key-query parameters. In this work, we establish a formal equivalence between the optimization geometry of self-attention and a hard-margin SVM problem that separates optimal input tokens from non-optimal tokens using linear constraints on the outer-products of token pa"},"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":"2308.16898","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-08-31T17:57:50Z","cross_cats_sorted":["cs.AI","cs.CL","math.OC"],"title_canon_sha256":"53f2e48176bb69dfff20c02461b5027e2c4f6196a900eb4a0fe41b351e735d90","abstract_canon_sha256":"7fe7e1c9113f996df9b04999579eacd44c18a7c47530bd070a369fe373889828"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:47:58.813401Z","signature_b64":"azeYachpuXiwmPRaeYdL8cU/3awa2qm1VavsmYBkrUM6Hit8GCv7xfjURbjlT+SEHs3ib9T8w70urvgyRiN8Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"62e2f673082c6a224b83d66a1b4deb119b86423808ec51625f95f01db78f7c43","last_reissued_at":"2026-07-05T07:47:58.812894Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:47:58.812894Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Transformers as Support Vector Machines","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","math.OC"],"primary_cat":"cs.LG","authors_text":"Christos Thrampoulidis, Davoud Ataee Tarzanagh, Samet Oymak, Yingcong Li","submitted_at":"2023-08-31T17:57:50Z","abstract_excerpt":"Since its inception in \"Attention Is All You Need\", transformer architecture has led to revolutionary advancements in NLP. The attention layer within the transformer admits a sequence of input tokens $X$ and makes them interact through pairwise similarities computed as softmax$(XQK^\\top X^\\top)$, where $(K,Q)$ are the trainable key-query parameters. In this work, we establish a formal equivalence between the optimization geometry of self-attention and a hard-margin SVM problem that separates optimal input tokens from non-optimal tokens using linear constraints on the outer-products of token pa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.16898","kind":"arxiv","version":3},"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/2308.16898/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":"2308.16898","created_at":"2026-07-05T07:47:58.812938+00:00"},{"alias_kind":"arxiv_version","alias_value":"2308.16898v3","created_at":"2026-07-05T07:47:58.812938+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.16898","created_at":"2026-07-05T07:47:58.812938+00:00"},{"alias_kind":"pith_short_12","alias_value":"MLRPM4YIFRVC","created_at":"2026-07-05T07:47:58.812938+00:00"},{"alias_kind":"pith_short_16","alias_value":"MLRPM4YIFRVCES4D","created_at":"2026-07-05T07:47:58.812938+00:00"},{"alias_kind":"pith_short_8","alias_value":"MLRPM4YI","created_at":"2026-07-05T07:47:58.812938+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12058","citing_title":"Phase Transitions in Attention: A Bayesian Theory of Copy Head Emergence","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09731","citing_title":"Tight Sample Complexity of Transformers","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23087","citing_title":"The Implicit Bias of Depth: From Neural Collapse to Softmax Codes","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10074","citing_title":"Transformers Learn the Optimal DDPM Denoiser for Multi-Token GMMs","ref_index":55,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MLRPM4YIFRVCES4D2ZVBWTPLCG","json":"https://pith.science/pith/MLRPM4YIFRVCES4D2ZVBWTPLCG.json","graph_json":"https://pith.science/api/pith-number/MLRPM4YIFRVCES4D2ZVBWTPLCG/graph.json","events_json":"https://pith.science/api/pith-number/MLRPM4YIFRVCES4D2ZVBWTPLCG/events.json","paper":"https://pith.science/paper/MLRPM4YI"},"agent_actions":{"view_html":"https://pith.science/pith/MLRPM4YIFRVCES4D2ZVBWTPLCG","download_json":"https://pith.science/pith/MLRPM4YIFRVCES4D2ZVBWTPLCG.json","view_paper":"https://pith.science/paper/MLRPM4YI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2308.16898&json=true","fetch_graph":"https://pith.science/api/pith-number/MLRPM4YIFRVCES4D2ZVBWTPLCG/graph.json","fetch_events":"https://pith.science/api/pith-number/MLRPM4YIFRVCES4D2ZVBWTPLCG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MLRPM4YIFRVCES4D2ZVBWTPLCG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MLRPM4YIFRVCES4D2ZVBWTPLCG/action/storage_attestation","attest_author":"https://pith.science/pith/MLRPM4YIFRVCES4D2ZVBWTPLCG/action/author_attestation","sign_citation":"https://pith.science/pith/MLRPM4YIFRVCES4D2ZVBWTPLCG/action/citation_signature","submit_replication":"https://pith.science/pith/MLRPM4YIFRVCES4D2ZVBWTPLCG/action/replication_record"}},"created_at":"2026-07-05T07:47:58.812938+00:00","updated_at":"2026-07-05T07:47:58.812938+00:00"}