{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4ZDYVDIJPFWLTIMK3D5NH55YGD","short_pith_number":"pith:4ZDYVDIJ","schema_version":"1.0","canonical_sha256":"e6478a8d09796cb9a18ad8fad3f7b830dcd355a70d9166b7dca143cf19236998","source":{"kind":"arxiv","id":"2410.09695","version":3},"attestation_state":"computed","paper":{"title":"Can In-context Learning Really Generalize to Out-of-distribution Tasks?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Qixun Wang, Xianghua Ying, Yifei Wang, Yisen Wang","submitted_at":"2024-10-13T02:10:26Z","abstract_excerpt":"In this work, we explore the mechanism of in-context learning (ICL) on out-of-distribution (OOD) tasks that were not encountered during training. To achieve this, we conduct synthetic experiments where the objective is to learn OOD mathematical functions through ICL using a GPT-2 model. We reveal that Transformers may struggle to learn OOD task functions through ICL. Specifically, ICL performance resembles implementing a function within the pretraining hypothesis space and optimizing it with gradient descent based on the in-context examples. Additionally, we investigate ICL's well-documented a"},"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":"2410.09695","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-13T02:10:26Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"104d78cd004b1044c92400f04e45b5f0b0a3a055a1be1a6da0bc0d52d41020c7","abstract_canon_sha256":"7be86d4403d21d9cd31e777e6e586b0a71919a7d7f354ca1f44333d9be6fe1be"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:44:15.506147Z","signature_b64":"ELGL8jBDzfJ4AcPxNzxugK2YOQgjE/g05dxGP+r9SNHvMGqun+OhNRuteLXflLmYubtimW8g4BMi6SlIOKTWAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e6478a8d09796cb9a18ad8fad3f7b830dcd355a70d9166b7dca143cf19236998","last_reissued_at":"2026-07-05T09:44:15.505582Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:44:15.505582Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Can In-context Learning Really Generalize to Out-of-distribution Tasks?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Qixun Wang, Xianghua Ying, Yifei Wang, Yisen Wang","submitted_at":"2024-10-13T02:10:26Z","abstract_excerpt":"In this work, we explore the mechanism of in-context learning (ICL) on out-of-distribution (OOD) tasks that were not encountered during training. To achieve this, we conduct synthetic experiments where the objective is to learn OOD mathematical functions through ICL using a GPT-2 model. We reveal that Transformers may struggle to learn OOD task functions through ICL. Specifically, ICL performance resembles implementing a function within the pretraining hypothesis space and optimizing it with gradient descent based on the in-context examples. Additionally, we investigate ICL's well-documented a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.09695","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/2410.09695/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":"2410.09695","created_at":"2026-07-05T09:44:15.505641+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.09695v3","created_at":"2026-07-05T09:44:15.505641+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.09695","created_at":"2026-07-05T09:44:15.505641+00:00"},{"alias_kind":"pith_short_12","alias_value":"4ZDYVDIJPFWL","created_at":"2026-07-05T09:44:15.505641+00:00"},{"alias_kind":"pith_short_16","alias_value":"4ZDYVDIJPFWLTIMK","created_at":"2026-07-05T09:44:15.505641+00:00"},{"alias_kind":"pith_short_8","alias_value":"4ZDYVDIJ","created_at":"2026-07-05T09:44:15.505641+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25450","citing_title":"The Generalization Spectrum: A Chromatographic Approach to Evaluating Learning Algorithms","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2606.25450","citing_title":"The Generalization Spectrum: A Chromatographic Approach to Evaluating Learning Algorithms","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2508.01191","citing_title":"Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4ZDYVDIJPFWLTIMK3D5NH55YGD","json":"https://pith.science/pith/4ZDYVDIJPFWLTIMK3D5NH55YGD.json","graph_json":"https://pith.science/api/pith-number/4ZDYVDIJPFWLTIMK3D5NH55YGD/graph.json","events_json":"https://pith.science/api/pith-number/4ZDYVDIJPFWLTIMK3D5NH55YGD/events.json","paper":"https://pith.science/paper/4ZDYVDIJ"},"agent_actions":{"view_html":"https://pith.science/pith/4ZDYVDIJPFWLTIMK3D5NH55YGD","download_json":"https://pith.science/pith/4ZDYVDIJPFWLTIMK3D5NH55YGD.json","view_paper":"https://pith.science/paper/4ZDYVDIJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.09695&json=true","fetch_graph":"https://pith.science/api/pith-number/4ZDYVDIJPFWLTIMK3D5NH55YGD/graph.json","fetch_events":"https://pith.science/api/pith-number/4ZDYVDIJPFWLTIMK3D5NH55YGD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4ZDYVDIJPFWLTIMK3D5NH55YGD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4ZDYVDIJPFWLTIMK3D5NH55YGD/action/storage_attestation","attest_author":"https://pith.science/pith/4ZDYVDIJPFWLTIMK3D5NH55YGD/action/author_attestation","sign_citation":"https://pith.science/pith/4ZDYVDIJPFWLTIMK3D5NH55YGD/action/citation_signature","submit_replication":"https://pith.science/pith/4ZDYVDIJPFWLTIMK3D5NH55YGD/action/replication_record"}},"created_at":"2026-07-05T09:44:15.505641+00:00","updated_at":"2026-07-05T09:44:15.505641+00:00"}