{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:57YZK25LAVNDA5775HL6CGZO5L","short_pith_number":"pith:57YZK25L","schema_version":"1.0","canonical_sha256":"eff1956bab055a3077ffe9d7e11b2eeac0283366b26aa7085e421d2a5e13fc2d","source":{"kind":"arxiv","id":"2106.03004","version":3},"attestation_state":"computed","paper":{"title":"Exploring the Limits of Out-of-Distribution Detection","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Balaji Lakshminarayanan, Jie Ren, Stanislav Fort","submitted_at":"2021-06-06T01:45:11Z","abstract_excerpt":"Near out-of-distribution detection (OOD) is a major challenge for deep neural networks. We demonstrate that large-scale pre-trained transformers can significantly improve the state-of-the-art (SOTA) on a range of near OOD tasks across different data modalities. For instance, on CIFAR-100 vs CIFAR-10 OOD detection, we improve the AUROC from 85% (current SOTA) to more than 96% using Vision Transformers pre-trained on ImageNet-21k. On a challenging genomics OOD detection benchmark, we improve the AUROC from 66% to 77% using transformers and unsupervised pre-training. To further improve performanc"},"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":"2106.03004","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-06-06T01:45:11Z","cross_cats_sorted":[],"title_canon_sha256":"c15713516cff3e297fe2884f59041c076a4f86c4c0f56272b03c4d94e43d67e8","abstract_canon_sha256":"51f071be67cc62072e84c96185ef307a4a20663270089b2bcdece28aced2c1de"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:01:36.426553Z","signature_b64":"b/KDLPhyp6Br/1YFrAsOBHY2Yorbl+BKlw4ssXxhSOqsByrcqhIj0R1Pm/EHzQKOYA/GqTxKlpvEtPK0qQGDBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eff1956bab055a3077ffe9d7e11b2eeac0283366b26aa7085e421d2a5e13fc2d","last_reissued_at":"2026-07-05T03:01:36.426202Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:01:36.426202Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Exploring the Limits of Out-of-Distribution Detection","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Balaji Lakshminarayanan, Jie Ren, Stanislav Fort","submitted_at":"2021-06-06T01:45:11Z","abstract_excerpt":"Near out-of-distribution detection (OOD) is a major challenge for deep neural networks. We demonstrate that large-scale pre-trained transformers can significantly improve the state-of-the-art (SOTA) on a range of near OOD tasks across different data modalities. For instance, on CIFAR-100 vs CIFAR-10 OOD detection, we improve the AUROC from 85% (current SOTA) to more than 96% using Vision Transformers pre-trained on ImageNet-21k. On a challenging genomics OOD detection benchmark, we improve the AUROC from 66% to 77% using transformers and unsupervised pre-training. To further improve performanc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.03004","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/2106.03004/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":"2106.03004","created_at":"2026-07-05T03:01:36.426259+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.03004v3","created_at":"2026-07-05T03:01:36.426259+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.03004","created_at":"2026-07-05T03:01:36.426259+00:00"},{"alias_kind":"pith_short_12","alias_value":"57YZK25LAVND","created_at":"2026-07-05T03:01:36.426259+00:00"},{"alias_kind":"pith_short_16","alias_value":"57YZK25LAVNDA577","created_at":"2026-07-05T03:01:36.426259+00:00"},{"alias_kind":"pith_short_8","alias_value":"57YZK25L","created_at":"2026-07-05T03:01:36.426259+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2303.08112","citing_title":"Eliciting Latent Predictions from Transformers with the Tuned Lens","ref_index":34,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/57YZK25LAVNDA5775HL6CGZO5L","json":"https://pith.science/pith/57YZK25LAVNDA5775HL6CGZO5L.json","graph_json":"https://pith.science/api/pith-number/57YZK25LAVNDA5775HL6CGZO5L/graph.json","events_json":"https://pith.science/api/pith-number/57YZK25LAVNDA5775HL6CGZO5L/events.json","paper":"https://pith.science/paper/57YZK25L"},"agent_actions":{"view_html":"https://pith.science/pith/57YZK25LAVNDA5775HL6CGZO5L","download_json":"https://pith.science/pith/57YZK25LAVNDA5775HL6CGZO5L.json","view_paper":"https://pith.science/paper/57YZK25L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.03004&json=true","fetch_graph":"https://pith.science/api/pith-number/57YZK25LAVNDA5775HL6CGZO5L/graph.json","fetch_events":"https://pith.science/api/pith-number/57YZK25LAVNDA5775HL6CGZO5L/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/57YZK25LAVNDA5775HL6CGZO5L/action/timestamp_anchor","attest_storage":"https://pith.science/pith/57YZK25LAVNDA5775HL6CGZO5L/action/storage_attestation","attest_author":"https://pith.science/pith/57YZK25LAVNDA5775HL6CGZO5L/action/author_attestation","sign_citation":"https://pith.science/pith/57YZK25LAVNDA5775HL6CGZO5L/action/citation_signature","submit_replication":"https://pith.science/pith/57YZK25LAVNDA5775HL6CGZO5L/action/replication_record"}},"created_at":"2026-07-05T03:01:36.426259+00:00","updated_at":"2026-07-05T03:01:36.426259+00:00"}