{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ACZNFCNIXLMHJ2L4HUHETXUUMF","short_pith_number":"pith:ACZNFCNI","schema_version":"1.0","canonical_sha256":"00b2d289a8bad874e97c3d0e49de94616f045511314a769ddabf0cf6883ef19a","source":{"kind":"arxiv","id":"2506.05200","version":2},"attestation_state":"computed","paper":{"title":"Transformers Meet In-Context Learning: A Universal Approximation Theory","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.ST","stat.ML","stat.TH"],"primary_cat":"cs.LG","authors_text":"Gen Li, Yuchen Jiao, Yu Huang, Yuting Wei, Yuxin Chen","submitted_at":"2025-06-05T16:12:51Z","abstract_excerpt":"Large language models are capable of in-context learning, the ability to perform new tasks at test time using a handful of input-output examples, without parameter updates. We develop a universal approximation theory to elucidate how transformers enable in-context learning. For a general class of functions (each representing a distinct task), we demonstrate how to construct a transformer that, without any further weight updates, can predict based on a few noisy in-context examples with vanishingly small risk. Unlike prior work that frames transformers as approximators of optimization algorithm"},"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":"2506.05200","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-05T16:12:51Z","cross_cats_sorted":["math.ST","stat.ML","stat.TH"],"title_canon_sha256":"9645b595940f769f2c9768b712421e5d6c9852441fcc859b7ff5155a8f6930d8","abstract_canon_sha256":"02fb5c6bd8ba1dc56e2e16cd37f176938a608d1658652ae96a842c3f8a2d0dc5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:00:50.496680Z","signature_b64":"qGqBemJXBA76pex7rEiKfrYyf7Yp3sdWpmQQ/b3JTRfWTD6Rjz0Jbm3S2Wq8yHMLBT/JhNTSpKFqkvWwNQZeBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"00b2d289a8bad874e97c3d0e49de94616f045511314a769ddabf0cf6883ef19a","last_reissued_at":"2026-07-05T12:00:50.496138Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:00:50.496138Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Transformers Meet In-Context Learning: A Universal Approximation Theory","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.ST","stat.ML","stat.TH"],"primary_cat":"cs.LG","authors_text":"Gen Li, Yuchen Jiao, Yu Huang, Yuting Wei, Yuxin Chen","submitted_at":"2025-06-05T16:12:51Z","abstract_excerpt":"Large language models are capable of in-context learning, the ability to perform new tasks at test time using a handful of input-output examples, without parameter updates. We develop a universal approximation theory to elucidate how transformers enable in-context learning. For a general class of functions (each representing a distinct task), we demonstrate how to construct a transformer that, without any further weight updates, can predict based on a few noisy in-context examples with vanishingly small risk. Unlike prior work that frames transformers as approximators of optimization algorithm"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.05200","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/2506.05200/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":"2506.05200","created_at":"2026-07-05T12:00:50.496203+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.05200v2","created_at":"2026-07-05T12:00:50.496203+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.05200","created_at":"2026-07-05T12:00:50.496203+00:00"},{"alias_kind":"pith_short_12","alias_value":"ACZNFCNIXLMH","created_at":"2026-07-05T12:00:50.496203+00:00"},{"alias_kind":"pith_short_16","alias_value":"ACZNFCNIXLMHJ2L4","created_at":"2026-07-05T12:00:50.496203+00:00"},{"alias_kind":"pith_short_8","alias_value":"ACZNFCNI","created_at":"2026-07-05T12:00:50.496203+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.05176","citing_title":"Understanding In-Context Learning for Nonlinear Regression with Transformers: Attention as Featurizer","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04995","citing_title":"Adaptivity Under Realizability Constraints: Comparing In-Context and Agentic Learning","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ACZNFCNIXLMHJ2L4HUHETXUUMF","json":"https://pith.science/pith/ACZNFCNIXLMHJ2L4HUHETXUUMF.json","graph_json":"https://pith.science/api/pith-number/ACZNFCNIXLMHJ2L4HUHETXUUMF/graph.json","events_json":"https://pith.science/api/pith-number/ACZNFCNIXLMHJ2L4HUHETXUUMF/events.json","paper":"https://pith.science/paper/ACZNFCNI"},"agent_actions":{"view_html":"https://pith.science/pith/ACZNFCNIXLMHJ2L4HUHETXUUMF","download_json":"https://pith.science/pith/ACZNFCNIXLMHJ2L4HUHETXUUMF.json","view_paper":"https://pith.science/paper/ACZNFCNI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.05200&json=true","fetch_graph":"https://pith.science/api/pith-number/ACZNFCNIXLMHJ2L4HUHETXUUMF/graph.json","fetch_events":"https://pith.science/api/pith-number/ACZNFCNIXLMHJ2L4HUHETXUUMF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ACZNFCNIXLMHJ2L4HUHETXUUMF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ACZNFCNIXLMHJ2L4HUHETXUUMF/action/storage_attestation","attest_author":"https://pith.science/pith/ACZNFCNIXLMHJ2L4HUHETXUUMF/action/author_attestation","sign_citation":"https://pith.science/pith/ACZNFCNIXLMHJ2L4HUHETXUUMF/action/citation_signature","submit_replication":"https://pith.science/pith/ACZNFCNIXLMHJ2L4HUHETXUUMF/action/replication_record"}},"created_at":"2026-07-05T12:00:50.496203+00:00","updated_at":"2026-07-05T12:00:50.496203+00:00"}