{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2016:IXRWK2OSQXW2NCCOM6CSRN462G","short_pith_number":"pith:IXRWK2OS","schema_version":"1.0","canonical_sha256":"45e36569d285eda6884e678528b79ed19ab4887641473d0b492d258377defbb0","source":{"kind":"arxiv","id":"1608.08614","version":2},"attestation_state":"computed","paper":{"title":"What makes ImageNet good for transfer learning?","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Alexei A. Efros, Minyoung Huh, Pulkit Agrawal","submitted_at":"2016-08-30T19:45:09Z","abstract_excerpt":"The tremendous success of ImageNet-trained deep features on a wide range of transfer tasks begs the question: what are the properties of the ImageNet dataset that are critical for learning good, general-purpose features? This work provides an empirical investigation of various facets of this question: Is more pre-training data always better? How does feature quality depend on the number of training examples per class? Does adding more object classes improve performance? For the same data budget, how should the data be split into classes? Is fine-grained recognition necessary for learning good "},"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":"1608.08614","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2016-08-30T19:45:09Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"6bb520e492794c3849aabe77d7f986974f2541bc138b4e75d8db827b4c2c4d79","abstract_canon_sha256":"49ca32bb61a64d2895b36930d305f27f6437b66debbb45fa40de723a542720f0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:55:20.741200Z","signature_b64":"XrvaeWs9QFJ2/D8iu3ucGjZSc9wsqo/KEXqD+UFGygzEFLxPZJuCacl9L7U74UWiHts7qsNhRtdxWntbXpNEBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"45e36569d285eda6884e678528b79ed19ab4887641473d0b492d258377defbb0","last_reissued_at":"2026-05-18T00:55:20.740669Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:55:20.740669Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What makes ImageNet good for transfer learning?","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Alexei A. Efros, Minyoung Huh, Pulkit Agrawal","submitted_at":"2016-08-30T19:45:09Z","abstract_excerpt":"The tremendous success of ImageNet-trained deep features on a wide range of transfer tasks begs the question: what are the properties of the ImageNet dataset that are critical for learning good, general-purpose features? This work provides an empirical investigation of various facets of this question: Is more pre-training data always better? How does feature quality depend on the number of training examples per class? Does adding more object classes improve performance? For the same data budget, how should the data be split into classes? Is fine-grained recognition necessary for learning good "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1608.08614","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":""},"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":"1608.08614","created_at":"2026-05-18T00:55:20.740750+00:00"},{"alias_kind":"arxiv_version","alias_value":"1608.08614v2","created_at":"2026-05-18T00:55:20.740750+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1608.08614","created_at":"2026-05-18T00:55:20.740750+00:00"},{"alias_kind":"pith_short_12","alias_value":"IXRWK2OSQXW2","created_at":"2026-05-18T12:30:22.444734+00:00"},{"alias_kind":"pith_short_16","alias_value":"IXRWK2OSQXW2NCCO","created_at":"2026-05-18T12:30:22.444734+00:00"},{"alias_kind":"pith_short_8","alias_value":"IXRWK2OS","created_at":"2026-05-18T12:30:22.444734+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"1907.07844","citing_title":"Growing a Brain: Fine-Tuning by Increasing Model Capacity","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2406.13187","citing_title":"Decouple then Converge: Handling Unknown Unlabeled Distributions in Long-Tailed Semi-Supervised Learning","ref_index":52,"is_internal_anchor":true},{"citing_arxiv_id":"2603.25476","citing_title":"How Class Ontology and Data Scale Affect Audio Transfer Learning","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"1910.04867","citing_title":"A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2604.27487","citing_title":"Low Rank Adaptation for Adversarial Perturbation","ref_index":70,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IXRWK2OSQXW2NCCOM6CSRN462G","json":"https://pith.science/pith/IXRWK2OSQXW2NCCOM6CSRN462G.json","graph_json":"https://pith.science/api/pith-number/IXRWK2OSQXW2NCCOM6CSRN462G/graph.json","events_json":"https://pith.science/api/pith-number/IXRWK2OSQXW2NCCOM6CSRN462G/events.json","paper":"https://pith.science/paper/IXRWK2OS"},"agent_actions":{"view_html":"https://pith.science/pith/IXRWK2OSQXW2NCCOM6CSRN462G","download_json":"https://pith.science/pith/IXRWK2OSQXW2NCCOM6CSRN462G.json","view_paper":"https://pith.science/paper/IXRWK2OS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1608.08614&json=true","fetch_graph":"https://pith.science/api/pith-number/IXRWK2OSQXW2NCCOM6CSRN462G/graph.json","fetch_events":"https://pith.science/api/pith-number/IXRWK2OSQXW2NCCOM6CSRN462G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IXRWK2OSQXW2NCCOM6CSRN462G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IXRWK2OSQXW2NCCOM6CSRN462G/action/storage_attestation","attest_author":"https://pith.science/pith/IXRWK2OSQXW2NCCOM6CSRN462G/action/author_attestation","sign_citation":"https://pith.science/pith/IXRWK2OSQXW2NCCOM6CSRN462G/action/citation_signature","submit_replication":"https://pith.science/pith/IXRWK2OSQXW2NCCOM6CSRN462G/action/replication_record"}},"created_at":"2026-05-18T00:55:20.740750+00:00","updated_at":"2026-05-18T00:55:20.740750+00:00"}