{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:522CHFUVK3PGKD6J6LVK7KBMZP","short_pith_number":"pith:522CHFUV","schema_version":"1.0","canonical_sha256":"eeb423969556de650fc9f2eaafa82ccbf0ae67e63eb494f045b41ffeb80fa107","source":{"kind":"arxiv","id":"2104.10972","version":4},"attestation_state":"computed","paper":{"title":"ImageNet-21K Pretraining for the Masses","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Asaf Noy, Emanuel Ben-Baruch, Lihi Zelnik-Manor, Tal Ridnik","submitted_at":"2021-04-22T10:10:14Z","abstract_excerpt":"ImageNet-1K serves as the primary dataset for pretraining deep learning models for computer vision tasks. ImageNet-21K dataset, which is bigger and more diverse, is used less frequently for pretraining, mainly due to its complexity, low accessibility, and underestimation of its added value. This paper aims to close this gap, and make high-quality efficient pretraining on ImageNet-21K available for everyone. Via a dedicated preprocessing stage, utilization of WordNet hierarchical structure, and a novel training scheme called semantic softmax, we show that various models significantly benefit fr"},"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":"2104.10972","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2021-04-22T10:10:14Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"80699c6dc579859b5e586fa8450bc315b05357716ccaaff69916b65875a25895","abstract_canon_sha256":"a5472805175b9ce6a5d8f7cc531645955a9d7c6bd0d416b195e55c78b1ab022a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:03:28.261264Z","signature_b64":"s/QrqlQfyR2qX9SqHg0B/Uc3Se7weL3OxdIlk0sJfo0HQD0BfnTNPJCi6HeYk79WfXF/bRdkzQqEIs26wpuDBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eeb423969556de650fc9f2eaafa82ccbf0ae67e63eb494f045b41ffeb80fa107","last_reissued_at":"2026-07-05T03:03:28.260849Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:03:28.260849Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ImageNet-21K Pretraining for the Masses","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Asaf Noy, Emanuel Ben-Baruch, Lihi Zelnik-Manor, Tal Ridnik","submitted_at":"2021-04-22T10:10:14Z","abstract_excerpt":"ImageNet-1K serves as the primary dataset for pretraining deep learning models for computer vision tasks. ImageNet-21K dataset, which is bigger and more diverse, is used less frequently for pretraining, mainly due to its complexity, low accessibility, and underestimation of its added value. This paper aims to close this gap, and make high-quality efficient pretraining on ImageNet-21K available for everyone. Via a dedicated preprocessing stage, utilization of WordNet hierarchical structure, and a novel training scheme called semantic softmax, we show that various models significantly benefit fr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2104.10972","kind":"arxiv","version":4},"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/2104.10972/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":"2104.10972","created_at":"2026-07-05T03:03:28.260910+00:00"},{"alias_kind":"arxiv_version","alias_value":"2104.10972v4","created_at":"2026-07-05T03:03:28.260910+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2104.10972","created_at":"2026-07-05T03:03:28.260910+00:00"},{"alias_kind":"pith_short_12","alias_value":"522CHFUVK3PG","created_at":"2026-07-05T03:03:28.260910+00:00"},{"alias_kind":"pith_short_16","alias_value":"522CHFUVK3PGKD6J","created_at":"2026-07-05T03:03:28.260910+00:00"},{"alias_kind":"pith_short_8","alias_value":"522CHFUV","created_at":"2026-07-05T03:03:28.260910+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":21,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09360","citing_title":"ExDet: Open-Domain Open-Vocabulary Detection with Cross-modal Extrapolation and Rectification","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00620","citing_title":"Identifying Latent Concepts and Structures for Generalized Category Discovery","ref_index":87,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00916","citing_title":"Condensing Large-Scale Datasets Directly with Minimal Information Loss","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02572","citing_title":"VISReg: Variance-Invariance-Sketching Regularization for JEPA training","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07558","citing_title":"Page image classifier fine-tuned on century-spanning archives of scanned documents for further content-specific processing","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27482","citing_title":"Energy-Structured Low-Rank Adaptation for Continual Learning","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26636","citing_title":"JetViT: Efficient High-Resolution Vision Transformer with Post-Training Attention Search","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28271","citing_title":"LV-OSD: Language-Vision-Complementary Open-Set Object Detection","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2606.16734","citing_title":"The Edge-on Galaxies in the DESI survey (EGIDE): sample building and photometry","ref_index":106,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23719","citing_title":"Weierstrass Positional Encoding for Vision Transformers","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2603.22570","citing_title":"CanViT: Toward Active-Vision Foundation Models","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2507.21114","citing_title":"Page image classification for content-specific data processing","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2511.22337","citing_title":"HandyLabel: Towards Post-Processing to Real-Time Annotation Using Skeleton Based Hand Gesture Recognition","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2310.01852","citing_title":"LanguageBind: Extending Video-Language Pretraining to N-modality by Language-based Semantic Alignment","ref_index":208,"is_internal_anchor":false},{"citing_arxiv_id":"2312.16886","citing_title":"MobileVLM : A Fast, Strong and Open Vision Language Assistant for Mobile Devices","ref_index":101,"is_internal_anchor":false},{"citing_arxiv_id":"2602.07940","citing_title":"MePo: Meta Post-Refinement for Rehearsal-Free General Continual Learning","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10953","citing_title":"Parameter-Efficient Adaptation of Pre-Trained Vision Foundation Models for Active and Passive Seismic Data Denoising","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11107","citing_title":"Birds of a Feather Flock Together: Background-Invariant Representations via Linear Structure in VLMs","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22477","citing_title":"Contrastive Semantic Projection: Faithful Neuron Labeling with Contrastive Examples","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13244","citing_title":"4th Workshop on Maritime Computer Vision (MaCVi): Challenge Overview","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04552","citing_title":"StableTTA: Improving Vision Model Performance by Training-free Test-Time Adaptation Methods","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/522CHFUVK3PGKD6J6LVK7KBMZP","json":"https://pith.science/pith/522CHFUVK3PGKD6J6LVK7KBMZP.json","graph_json":"https://pith.science/api/pith-number/522CHFUVK3PGKD6J6LVK7KBMZP/graph.json","events_json":"https://pith.science/api/pith-number/522CHFUVK3PGKD6J6LVK7KBMZP/events.json","paper":"https://pith.science/paper/522CHFUV"},"agent_actions":{"view_html":"https://pith.science/pith/522CHFUVK3PGKD6J6LVK7KBMZP","download_json":"https://pith.science/pith/522CHFUVK3PGKD6J6LVK7KBMZP.json","view_paper":"https://pith.science/paper/522CHFUV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2104.10972&json=true","fetch_graph":"https://pith.science/api/pith-number/522CHFUVK3PGKD6J6LVK7KBMZP/graph.json","fetch_events":"https://pith.science/api/pith-number/522CHFUVK3PGKD6J6LVK7KBMZP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/522CHFUVK3PGKD6J6LVK7KBMZP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/522CHFUVK3PGKD6J6LVK7KBMZP/action/storage_attestation","attest_author":"https://pith.science/pith/522CHFUVK3PGKD6J6LVK7KBMZP/action/author_attestation","sign_citation":"https://pith.science/pith/522CHFUVK3PGKD6J6LVK7KBMZP/action/citation_signature","submit_replication":"https://pith.science/pith/522CHFUVK3PGKD6J6LVK7KBMZP/action/replication_record"}},"created_at":"2026-07-05T03:03:28.260910+00:00","updated_at":"2026-07-05T03:03:28.260910+00:00"}