{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:UZAT3FNSWN7HH55YS2HTRRNHM5","short_pith_number":"pith:UZAT3FNS","schema_version":"1.0","canonical_sha256":"a6413d95b2b37e73f7b8968f38c5a767480648ee834145d593eb904433d76ade","source":{"kind":"arxiv","id":"2405.15618","version":3},"attestation_state":"computed","paper":{"title":"MLPs Learn In-Context on Regression and Classification Tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.NE"],"primary_cat":"cs.LG","authors_text":"Cengiz Pehlevan, William L. Tong","submitted_at":"2024-05-24T15:04:36Z","abstract_excerpt":"In-context learning (ICL), the remarkable ability to solve a task from only input exemplars, is often assumed to be a unique hallmark of Transformer models. By examining commonly employed synthetic ICL tasks, we demonstrate that multi-layer perceptrons (MLPs) can also learn in-context. Moreover, MLPs, and the closely related MLP-Mixer models, learn in-context comparably with Transformers under the same compute budget in this setting. We further show that MLPs outperform Transformers on a series of classical tasks from psychology designed to test relational reasoning, which are closely related "},"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":"2405.15618","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-24T15:04:36Z","cross_cats_sorted":["cs.NE"],"title_canon_sha256":"3a90f5a07ee2a7ba1c92c7374b5d592b6ab8607e809c594e170c8c4abf86995f","abstract_canon_sha256":"a42d546ba63a3cf71bb8c55d9093609260699424197ee64bd1f7836b54783139"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:19:24.387844Z","signature_b64":"L7kC8VWteF+CKrPjCbQAgMQqjcne5DurlssVjblVlhx1jQk5/9YIFMj3oGp4Rn5qQH3AmYV+BEDPu6uAwlm2Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a6413d95b2b37e73f7b8968f38c5a767480648ee834145d593eb904433d76ade","last_reissued_at":"2026-07-05T10:19:24.387307Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:19:24.387307Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MLPs Learn In-Context on Regression and Classification Tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.NE"],"primary_cat":"cs.LG","authors_text":"Cengiz Pehlevan, William L. Tong","submitted_at":"2024-05-24T15:04:36Z","abstract_excerpt":"In-context learning (ICL), the remarkable ability to solve a task from only input exemplars, is often assumed to be a unique hallmark of Transformer models. By examining commonly employed synthetic ICL tasks, we demonstrate that multi-layer perceptrons (MLPs) can also learn in-context. Moreover, MLPs, and the closely related MLP-Mixer models, learn in-context comparably with Transformers under the same compute budget in this setting. We further show that MLPs outperform Transformers on a series of classical tasks from psychology designed to test relational reasoning, which are closely related "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.15618","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/2405.15618/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":"2405.15618","created_at":"2026-07-05T10:19:24.387385+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.15618v3","created_at":"2026-07-05T10:19:24.387385+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.15618","created_at":"2026-07-05T10:19:24.387385+00:00"},{"alias_kind":"pith_short_12","alias_value":"UZAT3FNSWN7H","created_at":"2026-07-05T10:19:24.387385+00:00"},{"alias_kind":"pith_short_16","alias_value":"UZAT3FNSWN7HH55Y","created_at":"2026-07-05T10:19:24.387385+00:00"},{"alias_kind":"pith_short_8","alias_value":"UZAT3FNS","created_at":"2026-07-05T10:19:24.387385+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2508.04461","citing_title":"Small transformer architectures for task switching","ref_index":21,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UZAT3FNSWN7HH55YS2HTRRNHM5","json":"https://pith.science/pith/UZAT3FNSWN7HH55YS2HTRRNHM5.json","graph_json":"https://pith.science/api/pith-number/UZAT3FNSWN7HH55YS2HTRRNHM5/graph.json","events_json":"https://pith.science/api/pith-number/UZAT3FNSWN7HH55YS2HTRRNHM5/events.json","paper":"https://pith.science/paper/UZAT3FNS"},"agent_actions":{"view_html":"https://pith.science/pith/UZAT3FNSWN7HH55YS2HTRRNHM5","download_json":"https://pith.science/pith/UZAT3FNSWN7HH55YS2HTRRNHM5.json","view_paper":"https://pith.science/paper/UZAT3FNS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.15618&json=true","fetch_graph":"https://pith.science/api/pith-number/UZAT3FNSWN7HH55YS2HTRRNHM5/graph.json","fetch_events":"https://pith.science/api/pith-number/UZAT3FNSWN7HH55YS2HTRRNHM5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UZAT3FNSWN7HH55YS2HTRRNHM5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UZAT3FNSWN7HH55YS2HTRRNHM5/action/storage_attestation","attest_author":"https://pith.science/pith/UZAT3FNSWN7HH55YS2HTRRNHM5/action/author_attestation","sign_citation":"https://pith.science/pith/UZAT3FNSWN7HH55YS2HTRRNHM5/action/citation_signature","submit_replication":"https://pith.science/pith/UZAT3FNSWN7HH55YS2HTRRNHM5/action/replication_record"}},"created_at":"2026-07-05T10:19:24.387385+00:00","updated_at":"2026-07-05T10:19:24.387385+00:00"}