{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:OTNFQZJLWMAU7G4SRJQYMKQALR","short_pith_number":"pith:OTNFQZJL","schema_version":"1.0","canonical_sha256":"74da58652bb3014f9b928a61862a005c64d0dabafc7ba4ee2c3d3fd51a35068a","source":{"kind":"arxiv","id":"2504.05461","version":1},"attestation_state":"computed","paper":{"title":"Intermediate Layer Classifiers for OOD generalization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Arnas Uselis, Seong Joon Oh","submitted_at":"2025-04-07T19:50:50Z","abstract_excerpt":"Deep classifiers are known to be sensitive to data distribution shifts, primarily due to their reliance on spurious correlations in training data. It has been suggested that these classifiers can still find useful features in the network's last layer that hold up under such shifts. In this work, we question the use of last-layer representations for out-of-distribution (OOD) generalisation and explore the utility of intermediate layers. To this end, we introduce \\textit{Intermediate Layer Classifiers} (ILCs). We discover that intermediate layer representations frequently offer substantially bet"},"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":"2504.05461","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-07T19:50:50Z","cross_cats_sorted":[],"title_canon_sha256":"fb73485cec56bf5cabba43ead5a9f5bfa238787ce322d2263c4cc446e1a59b1f","abstract_canon_sha256":"4592aea178092e17f8e8708f42d7f8417f7bd9a986af2798e3db8e8a4ff64ef7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:45:42.777375Z","signature_b64":"V83d7oLolAigxaLwTofBNZJdx7kL2dTy8wFC8PS86w7LWLfoV4FEQOuPnonUXRI9fZU9BagSVxmLf+XE+qoKBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"74da58652bb3014f9b928a61862a005c64d0dabafc7ba4ee2c3d3fd51a35068a","last_reissued_at":"2026-07-05T10:45:42.776919Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:45:42.776919Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Intermediate Layer Classifiers for OOD generalization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Arnas Uselis, Seong Joon Oh","submitted_at":"2025-04-07T19:50:50Z","abstract_excerpt":"Deep classifiers are known to be sensitive to data distribution shifts, primarily due to their reliance on spurious correlations in training data. It has been suggested that these classifiers can still find useful features in the network's last layer that hold up under such shifts. In this work, we question the use of last-layer representations for out-of-distribution (OOD) generalisation and explore the utility of intermediate layers. To this end, we introduce \\textit{Intermediate Layer Classifiers} (ILCs). We discover that intermediate layer representations frequently offer substantially bet"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.05461","kind":"arxiv","version":1},"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/2504.05461/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":"2504.05461","created_at":"2026-07-05T10:45:42.776975+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.05461v1","created_at":"2026-07-05T10:45:42.776975+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.05461","created_at":"2026-07-05T10:45:42.776975+00:00"},{"alias_kind":"pith_short_12","alias_value":"OTNFQZJLWMAU","created_at":"2026-07-05T10:45:42.776975+00:00"},{"alias_kind":"pith_short_16","alias_value":"OTNFQZJLWMAU7G4S","created_at":"2026-07-05T10:45:42.776975+00:00"},{"alias_kind":"pith_short_8","alias_value":"OTNFQZJL","created_at":"2026-07-05T10:45:42.776975+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.05387","citing_title":"The Generalization Ridge: Information Flow in Natural Language Generation","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OTNFQZJLWMAU7G4SRJQYMKQALR","json":"https://pith.science/pith/OTNFQZJLWMAU7G4SRJQYMKQALR.json","graph_json":"https://pith.science/api/pith-number/OTNFQZJLWMAU7G4SRJQYMKQALR/graph.json","events_json":"https://pith.science/api/pith-number/OTNFQZJLWMAU7G4SRJQYMKQALR/events.json","paper":"https://pith.science/paper/OTNFQZJL"},"agent_actions":{"view_html":"https://pith.science/pith/OTNFQZJLWMAU7G4SRJQYMKQALR","download_json":"https://pith.science/pith/OTNFQZJLWMAU7G4SRJQYMKQALR.json","view_paper":"https://pith.science/paper/OTNFQZJL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.05461&json=true","fetch_graph":"https://pith.science/api/pith-number/OTNFQZJLWMAU7G4SRJQYMKQALR/graph.json","fetch_events":"https://pith.science/api/pith-number/OTNFQZJLWMAU7G4SRJQYMKQALR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OTNFQZJLWMAU7G4SRJQYMKQALR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OTNFQZJLWMAU7G4SRJQYMKQALR/action/storage_attestation","attest_author":"https://pith.science/pith/OTNFQZJLWMAU7G4SRJQYMKQALR/action/author_attestation","sign_citation":"https://pith.science/pith/OTNFQZJLWMAU7G4SRJQYMKQALR/action/citation_signature","submit_replication":"https://pith.science/pith/OTNFQZJLWMAU7G4SRJQYMKQALR/action/replication_record"}},"created_at":"2026-07-05T10:45:42.776975+00:00","updated_at":"2026-07-05T10:45:42.776975+00:00"}