{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LPKFDEF2KUDOM5LKICLATM7EFQ","short_pith_number":"pith:LPKFDEF2","schema_version":"1.0","canonical_sha256":"5bd45190ba5506e6756a409609b3e42c29828d6507c2972dd86c7ed167ca72c5","source":{"kind":"arxiv","id":"2502.03327","version":1},"attestation_state":"computed","paper":{"title":"Is In-Context Universality Enough? MLPs are Also Universal In-Context","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","cs.NA","cs.NE","math.NA","math.PR"],"primary_cat":"stat.ML","authors_text":"Anastasis Kratsios, Takashi Furuya","submitted_at":"2025-02-05T16:22:46Z","abstract_excerpt":"The success of transformers is often linked to their ability to perform in-context learning. Recent work shows that transformers are universal in context, capable of approximating any real-valued continuous function of a context (a probability measure over $\\mathcal{X}\\subseteq \\mathbb{R}^d$) and a query $x\\in \\mathcal{X}$. This raises the question: Does in-context universality explain their advantage over classical models? We answer this in the negative by proving that MLPs with trainable activation functions are also universal in-context. This suggests the transformer's success is likely due"},"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":"2502.03327","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-02-05T16:22:46Z","cross_cats_sorted":["cs.LG","cs.NA","cs.NE","math.NA","math.PR"],"title_canon_sha256":"5f5b904a0e9edd13ba48fbae4804b33fcbafc8a2a2e762404ddcfb6342e1b09f","abstract_canon_sha256":"806bf0ea619aad41f547e2fe585fd6d46076305293a1b81de068801f057faa85"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:10:03.258535Z","signature_b64":"RtFYeGHPQ+6nbAjsYurZYS/hGyhdouICXfl/iFaJJwEkbV3xeb2HeIw2yyyCEFwO+PI2MbvATk3vyYsRoLNiDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5bd45190ba5506e6756a409609b3e42c29828d6507c2972dd86c7ed167ca72c5","last_reissued_at":"2026-07-05T10:10:03.258057Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:10:03.258057Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Is In-Context Universality Enough? MLPs are Also Universal In-Context","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","cs.NA","cs.NE","math.NA","math.PR"],"primary_cat":"stat.ML","authors_text":"Anastasis Kratsios, Takashi Furuya","submitted_at":"2025-02-05T16:22:46Z","abstract_excerpt":"The success of transformers is often linked to their ability to perform in-context learning. Recent work shows that transformers are universal in context, capable of approximating any real-valued continuous function of a context (a probability measure over $\\mathcal{X}\\subseteq \\mathbb{R}^d$) and a query $x\\in \\mathcal{X}$. This raises the question: Does in-context universality explain their advantage over classical models? We answer this in the negative by proving that MLPs with trainable activation functions are also universal in-context. This suggests the transformer's success is likely due"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.03327","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/2502.03327/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":"2502.03327","created_at":"2026-07-05T10:10:03.258114+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.03327v1","created_at":"2026-07-05T10:10:03.258114+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.03327","created_at":"2026-07-05T10:10:03.258114+00:00"},{"alias_kind":"pith_short_12","alias_value":"LPKFDEF2KUDO","created_at":"2026-07-05T10:10:03.258114+00:00"},{"alias_kind":"pith_short_16","alias_value":"LPKFDEF2KUDOM5LK","created_at":"2026-07-05T10:10:03.258114+00:00"},{"alias_kind":"pith_short_8","alias_value":"LPKFDEF2","created_at":"2026-07-05T10:10:03.258114+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.04995","citing_title":"Adaptivity Under Realizability Constraints: Comparing In-Context and Agentic Learning","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LPKFDEF2KUDOM5LKICLATM7EFQ","json":"https://pith.science/pith/LPKFDEF2KUDOM5LKICLATM7EFQ.json","graph_json":"https://pith.science/api/pith-number/LPKFDEF2KUDOM5LKICLATM7EFQ/graph.json","events_json":"https://pith.science/api/pith-number/LPKFDEF2KUDOM5LKICLATM7EFQ/events.json","paper":"https://pith.science/paper/LPKFDEF2"},"agent_actions":{"view_html":"https://pith.science/pith/LPKFDEF2KUDOM5LKICLATM7EFQ","download_json":"https://pith.science/pith/LPKFDEF2KUDOM5LKICLATM7EFQ.json","view_paper":"https://pith.science/paper/LPKFDEF2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.03327&json=true","fetch_graph":"https://pith.science/api/pith-number/LPKFDEF2KUDOM5LKICLATM7EFQ/graph.json","fetch_events":"https://pith.science/api/pith-number/LPKFDEF2KUDOM5LKICLATM7EFQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LPKFDEF2KUDOM5LKICLATM7EFQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LPKFDEF2KUDOM5LKICLATM7EFQ/action/storage_attestation","attest_author":"https://pith.science/pith/LPKFDEF2KUDOM5LKICLATM7EFQ/action/author_attestation","sign_citation":"https://pith.science/pith/LPKFDEF2KUDOM5LKICLATM7EFQ/action/citation_signature","submit_replication":"https://pith.science/pith/LPKFDEF2KUDOM5LKICLATM7EFQ/action/replication_record"}},"created_at":"2026-07-05T10:10:03.258114+00:00","updated_at":"2026-07-05T10:10:03.258114+00:00"}