{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:35RY7WO4A4KEWZUZMB5ODGIPBL","short_pith_number":"pith:35RY7WO4","schema_version":"1.0","canonical_sha256":"df638fd9dc07144b6699607ae1990f0adc96314f63f616841ef808d3fcc1998e","source":{"kind":"arxiv","id":"2501.01312","version":3},"attestation_state":"computed","paper":{"title":"Learning Spectral Methods by Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","math.ST","stat.TH"],"primary_cat":"stat.ML","authors_text":"Dennis Wu, Han Liu, Hong-Yu Chen, Jianqing Fan, Yihan He, Yuan Cao","submitted_at":"2025-01-02T15:53:25Z","abstract_excerpt":"Transformers demonstrate significant advantages as the building block of modern LLMs. In this work, we study the capacities of Transformers in performing unsupervised learning. We show that multi-layered Transformers, given a sufficiently large set of pre-training instances, are able to learn the algorithms themselves and perform statistical estimation tasks given new instances. This learning paradigm is distinct from the in-context learning setup and is similar to the learning procedure of human brains where skills are learned through past experience. Theoretically, we prove that pre-trained "},"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":"2501.01312","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-01-02T15:53:25Z","cross_cats_sorted":["cs.LG","math.ST","stat.TH"],"title_canon_sha256":"d742706719db643bc70ca297abee950b5a9a74cccfe158364854f61b677a5d3b","abstract_canon_sha256":"670d61915af1c6a37676ffd42912ed758525f7f4a36707efaa5d6b6552ea218e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:59:58.389651Z","signature_b64":"m06oCVXd4ow8/IVKUEFFpHl3C24cn9GiX712L17N3dlW1KLePBoU/AswBycU2p1vITtHY3BtGckYuU8stO/rAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"df638fd9dc07144b6699607ae1990f0adc96314f63f616841ef808d3fcc1998e","last_reissued_at":"2026-07-05T09:59:58.389205Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:59:58.389205Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Spectral Methods by Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","math.ST","stat.TH"],"primary_cat":"stat.ML","authors_text":"Dennis Wu, Han Liu, Hong-Yu Chen, Jianqing Fan, Yihan He, Yuan Cao","submitted_at":"2025-01-02T15:53:25Z","abstract_excerpt":"Transformers demonstrate significant advantages as the building block of modern LLMs. In this work, we study the capacities of Transformers in performing unsupervised learning. We show that multi-layered Transformers, given a sufficiently large set of pre-training instances, are able to learn the algorithms themselves and perform statistical estimation tasks given new instances. This learning paradigm is distinct from the in-context learning setup and is similar to the learning procedure of human brains where skills are learned through past experience. Theoretically, we prove that pre-trained "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.01312","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/2501.01312/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":"2501.01312","created_at":"2026-07-05T09:59:58.389257+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.01312v3","created_at":"2026-07-05T09:59:58.389257+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.01312","created_at":"2026-07-05T09:59:58.389257+00:00"},{"alias_kind":"pith_short_12","alias_value":"35RY7WO4A4KE","created_at":"2026-07-05T09:59:58.389257+00:00"},{"alias_kind":"pith_short_16","alias_value":"35RY7WO4A4KEWZUZ","created_at":"2026-07-05T09:59:58.389257+00:00"},{"alias_kind":"pith_short_8","alias_value":"35RY7WO4","created_at":"2026-07-05T09:59:58.389257+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.06609","citing_title":"Transformers Efficiently Perform In-Context Logistic Regression via Normalized Gradient Descent","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/35RY7WO4A4KEWZUZMB5ODGIPBL","json":"https://pith.science/pith/35RY7WO4A4KEWZUZMB5ODGIPBL.json","graph_json":"https://pith.science/api/pith-number/35RY7WO4A4KEWZUZMB5ODGIPBL/graph.json","events_json":"https://pith.science/api/pith-number/35RY7WO4A4KEWZUZMB5ODGIPBL/events.json","paper":"https://pith.science/paper/35RY7WO4"},"agent_actions":{"view_html":"https://pith.science/pith/35RY7WO4A4KEWZUZMB5ODGIPBL","download_json":"https://pith.science/pith/35RY7WO4A4KEWZUZMB5ODGIPBL.json","view_paper":"https://pith.science/paper/35RY7WO4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.01312&json=true","fetch_graph":"https://pith.science/api/pith-number/35RY7WO4A4KEWZUZMB5ODGIPBL/graph.json","fetch_events":"https://pith.science/api/pith-number/35RY7WO4A4KEWZUZMB5ODGIPBL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/35RY7WO4A4KEWZUZMB5ODGIPBL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/35RY7WO4A4KEWZUZMB5ODGIPBL/action/storage_attestation","attest_author":"https://pith.science/pith/35RY7WO4A4KEWZUZMB5ODGIPBL/action/author_attestation","sign_citation":"https://pith.science/pith/35RY7WO4A4KEWZUZMB5ODGIPBL/action/citation_signature","submit_replication":"https://pith.science/pith/35RY7WO4A4KEWZUZMB5ODGIPBL/action/replication_record"}},"created_at":"2026-07-05T09:59:58.389257+00:00","updated_at":"2026-07-05T09:59:58.389257+00:00"}