{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:HI53VTSLMZ6G3AZCJO6JHD2YAF","short_pith_number":"pith:HI53VTSL","schema_version":"1.0","canonical_sha256":"3a3bbace4b667c6d83224bbc938f58015899c08cd6c737b078dce2555af0f2a6","source":{"kind":"arxiv","id":"2301.07067","version":2},"attestation_state":"computed","paper":{"title":"Transformers as Algorithms: Generalization and Stability in In-context Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Dimitris Papailiopoulos, M. Emrullah Ildiz, Samet Oymak, Yingcong Li","submitted_at":"2023-01-17T18:31:12Z","abstract_excerpt":"In-context learning (ICL) is a type of prompting where a transformer model operates on a sequence of (input, output) examples and performs inference on-the-fly. In this work, we formalize in-context learning as an algorithm learning problem where a transformer model implicitly constructs a hypothesis function at inference-time. We first explore the statistical aspects of this abstraction through the lens of multitask learning: We obtain generalization bounds for ICL when the input prompt is (1) a sequence of i.i.d. (input, label) pairs or (2) a trajectory arising from a dynamical system. The c"},"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":"2301.07067","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-01-17T18:31:12Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"3ca5397fcb73b78bee0563493b39e965a61c70bdce51ad1b88170bd44a4933a1","abstract_canon_sha256":"124efadf65694940eed8584dace55b2b9140c008bba2042c2a50fb4d57b79140"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:39:04.406412Z","signature_b64":"eaiv7qzySVZ+xFh+x1B+l4c6yGbH/kc5hnMhTEpUs2Or7vWZeRQgUV+LyK+bMuOBDnaTo1GiGDg0eHLGVkgXAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3a3bbace4b667c6d83224bbc938f58015899c08cd6c737b078dce2555af0f2a6","last_reissued_at":"2026-07-05T05:39:04.405895Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:39:04.405895Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Transformers as Algorithms: Generalization and Stability in In-context Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Dimitris Papailiopoulos, M. Emrullah Ildiz, Samet Oymak, Yingcong Li","submitted_at":"2023-01-17T18:31:12Z","abstract_excerpt":"In-context learning (ICL) is a type of prompting where a transformer model operates on a sequence of (input, output) examples and performs inference on-the-fly. In this work, we formalize in-context learning as an algorithm learning problem where a transformer model implicitly constructs a hypothesis function at inference-time. We first explore the statistical aspects of this abstraction through the lens of multitask learning: We obtain generalization bounds for ICL when the input prompt is (1) a sequence of i.i.d. (input, label) pairs or (2) a trajectory arising from a dynamical system. The c"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2301.07067","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/2301.07067/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":"2301.07067","created_at":"2026-07-05T05:39:04.405955+00:00"},{"alias_kind":"arxiv_version","alias_value":"2301.07067v2","created_at":"2026-07-05T05:39:04.405955+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2301.07067","created_at":"2026-07-05T05:39:04.405955+00:00"},{"alias_kind":"pith_short_12","alias_value":"HI53VTSLMZ6G","created_at":"2026-07-05T05:39:04.405955+00:00"},{"alias_kind":"pith_short_16","alias_value":"HI53VTSLMZ6G3AZC","created_at":"2026-07-05T05:39:04.405955+00:00"},{"alias_kind":"pith_short_8","alias_value":"HI53VTSL","created_at":"2026-07-05T05:39:04.405955+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.02211","citing_title":"Consistency Training while Mitigating Obfuscation via Rate Matching","ref_index":118,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HI53VTSLMZ6G3AZCJO6JHD2YAF","json":"https://pith.science/pith/HI53VTSLMZ6G3AZCJO6JHD2YAF.json","graph_json":"https://pith.science/api/pith-number/HI53VTSLMZ6G3AZCJO6JHD2YAF/graph.json","events_json":"https://pith.science/api/pith-number/HI53VTSLMZ6G3AZCJO6JHD2YAF/events.json","paper":"https://pith.science/paper/HI53VTSL"},"agent_actions":{"view_html":"https://pith.science/pith/HI53VTSLMZ6G3AZCJO6JHD2YAF","download_json":"https://pith.science/pith/HI53VTSLMZ6G3AZCJO6JHD2YAF.json","view_paper":"https://pith.science/paper/HI53VTSL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2301.07067&json=true","fetch_graph":"https://pith.science/api/pith-number/HI53VTSLMZ6G3AZCJO6JHD2YAF/graph.json","fetch_events":"https://pith.science/api/pith-number/HI53VTSLMZ6G3AZCJO6JHD2YAF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HI53VTSLMZ6G3AZCJO6JHD2YAF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HI53VTSLMZ6G3AZCJO6JHD2YAF/action/storage_attestation","attest_author":"https://pith.science/pith/HI53VTSLMZ6G3AZCJO6JHD2YAF/action/author_attestation","sign_citation":"https://pith.science/pith/HI53VTSLMZ6G3AZCJO6JHD2YAF/action/citation_signature","submit_replication":"https://pith.science/pith/HI53VTSLMZ6G3AZCJO6JHD2YAF/action/replication_record"}},"created_at":"2026-07-05T05:39:04.405955+00:00","updated_at":"2026-07-05T05:39:04.405955+00:00"}