{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:VUNJYKIRF4ZEWXZ33JNOTRELRK","short_pith_number":"pith:VUNJYKIR","schema_version":"1.0","canonical_sha256":"ad1a9c29112f324b5f3bda5ae9c48b8ab90c8a63f2e42ab53e5369a3cb8dde5a","source":{"kind":"arxiv","id":"2306.15769","version":2},"attestation_state":"computed","paper":{"title":"What Makes ImageNet Look Unlike LAION","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Ali Shirali, Moritz Hardt","submitted_at":"2023-06-27T19:34:53Z","abstract_excerpt":"ImageNet was famously created from Flickr image search results. What if we recreated ImageNet instead by searching the massive LAION dataset based on image captions alone? In this work, we carry out this counterfactual investigation. We find that the resulting ImageNet recreation, which we call LAIONet, looks distinctly unlike the original. Specifically, the intra-class similarity of images in the original ImageNet is dramatically higher than it is for LAIONet. Consequently, models trained on ImageNet perform significantly worse on LAIONet. We propose a rigorous explanation for the discrepancy"},"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":"2306.15769","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-06-27T19:34:53Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"cbe8bb9943d654220b3c0495afa21cab384bd947c3921a6791fe365be13d1bf1","abstract_canon_sha256":"9fdef583489d3a30954fd195e1cefd7465c336198a7c8da68a105d61367b587a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:27:51.082734Z","signature_b64":"jVZjpTjTMX69AkhySh9EWXP/hxuSpvO2p1ozVWYxMClDvmtf2VgGWKWqrz0xuOoRXL0iSwmsUcsEArefUMKwDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ad1a9c29112f324b5f3bda5ae9c48b8ab90c8a63f2e42ab53e5369a3cb8dde5a","last_reissued_at":"2026-07-05T09:27:51.082269Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:27:51.082269Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What Makes ImageNet Look Unlike LAION","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Ali Shirali, Moritz Hardt","submitted_at":"2023-06-27T19:34:53Z","abstract_excerpt":"ImageNet was famously created from Flickr image search results. What if we recreated ImageNet instead by searching the massive LAION dataset based on image captions alone? In this work, we carry out this counterfactual investigation. We find that the resulting ImageNet recreation, which we call LAIONet, looks distinctly unlike the original. Specifically, the intra-class similarity of images in the original ImageNet is dramatically higher than it is for LAIONet. Consequently, models trained on ImageNet perform significantly worse on LAIONet. We propose a rigorous explanation for the discrepancy"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.15769","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/2306.15769/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":"2306.15769","created_at":"2026-07-05T09:27:51.082333+00:00"},{"alias_kind":"arxiv_version","alias_value":"2306.15769v2","created_at":"2026-07-05T09:27:51.082333+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.15769","created_at":"2026-07-05T09:27:51.082333+00:00"},{"alias_kind":"pith_short_12","alias_value":"VUNJYKIRF4ZE","created_at":"2026-07-05T09:27:51.082333+00:00"},{"alias_kind":"pith_short_16","alias_value":"VUNJYKIRF4ZEWXZ3","created_at":"2026-07-05T09:27:51.082333+00:00"},{"alias_kind":"pith_short_8","alias_value":"VUNJYKIR","created_at":"2026-07-05T09:27:51.082333+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.14186","citing_title":"Setting-Matched and Semantics-Scaled Benchmarking of One-Step Generative Models Against Multistep Diffusion and Flow Models","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VUNJYKIRF4ZEWXZ33JNOTRELRK","json":"https://pith.science/pith/VUNJYKIRF4ZEWXZ33JNOTRELRK.json","graph_json":"https://pith.science/api/pith-number/VUNJYKIRF4ZEWXZ33JNOTRELRK/graph.json","events_json":"https://pith.science/api/pith-number/VUNJYKIRF4ZEWXZ33JNOTRELRK/events.json","paper":"https://pith.science/paper/VUNJYKIR"},"agent_actions":{"view_html":"https://pith.science/pith/VUNJYKIRF4ZEWXZ33JNOTRELRK","download_json":"https://pith.science/pith/VUNJYKIRF4ZEWXZ33JNOTRELRK.json","view_paper":"https://pith.science/paper/VUNJYKIR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2306.15769&json=true","fetch_graph":"https://pith.science/api/pith-number/VUNJYKIRF4ZEWXZ33JNOTRELRK/graph.json","fetch_events":"https://pith.science/api/pith-number/VUNJYKIRF4ZEWXZ33JNOTRELRK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VUNJYKIRF4ZEWXZ33JNOTRELRK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VUNJYKIRF4ZEWXZ33JNOTRELRK/action/storage_attestation","attest_author":"https://pith.science/pith/VUNJYKIRF4ZEWXZ33JNOTRELRK/action/author_attestation","sign_citation":"https://pith.science/pith/VUNJYKIRF4ZEWXZ33JNOTRELRK/action/citation_signature","submit_replication":"https://pith.science/pith/VUNJYKIRF4ZEWXZ33JNOTRELRK/action/replication_record"}},"created_at":"2026-07-05T09:27:51.082333+00:00","updated_at":"2026-07-05T09:27:51.082333+00:00"}