{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:BHAUKIF434T4PD6UWSKNTDIOD5","short_pith_number":"pith:BHAUKIF4","schema_version":"1.0","canonical_sha256":"09c14520bcdf27c78fd4b494d98d0e1f45537db8ac58428c997ce91aa10f1437","source":{"kind":"arxiv","id":"2202.06523","version":1},"attestation_state":"computed","paper":{"title":"MetaShift: A Dataset of Datasets for Evaluating Contextual Distribution Shifts and Training Conflicts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"James Zou, Weixin Liang","submitted_at":"2022-02-14T07:40:03Z","abstract_excerpt":"Understanding the performance of machine learning models across diverse data distributions is critically important for reliable applications. Motivated by this, there is a growing focus on curating benchmark datasets that capture distribution shifts. While valuable, the existing benchmarks are limited in that many of them only contain a small number of shifts and they lack systematic annotation about what is different across different shifts. We present MetaShift--a collection of 12,868 sets of natural images across 410 classes--to address this challenge. We leverage the natural heterogeneity "},"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":"2202.06523","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-02-14T07:40:03Z","cross_cats_sorted":["cs.AI","cs.CL","cs.CV"],"title_canon_sha256":"46714a9fe939024785ce6e2cd1f7c373fcebdafc253acbec8b84732341d4e221","abstract_canon_sha256":"d63e9162b61cba821e4441d73fe2b10eb48bfd2217d2580d3048dbd095226a06"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:56:39.462776Z","signature_b64":"KrqF1l7R0feVMhLu8pqQcydpjVZQgJVaQA4ZbdeFIHy/D3vdAJYZm8QoQuDnvTmHLwR/sJgAvIEbf/TyI2QuCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"09c14520bcdf27c78fd4b494d98d0e1f45537db8ac58428c997ce91aa10f1437","last_reissued_at":"2026-07-05T03:56:39.462350Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:56:39.462350Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MetaShift: A Dataset of Datasets for Evaluating Contextual Distribution Shifts and Training Conflicts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"James Zou, Weixin Liang","submitted_at":"2022-02-14T07:40:03Z","abstract_excerpt":"Understanding the performance of machine learning models across diverse data distributions is critically important for reliable applications. Motivated by this, there is a growing focus on curating benchmark datasets that capture distribution shifts. While valuable, the existing benchmarks are limited in that many of them only contain a small number of shifts and they lack systematic annotation about what is different across different shifts. We present MetaShift--a collection of 12,868 sets of natural images across 410 classes--to address this challenge. We leverage the natural heterogeneity "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.06523","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/2202.06523/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":"2202.06523","created_at":"2026-07-05T03:56:39.462409+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.06523v1","created_at":"2026-07-05T03:56:39.462409+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.06523","created_at":"2026-07-05T03:56:39.462409+00:00"},{"alias_kind":"pith_short_12","alias_value":"BHAUKIF434T4","created_at":"2026-07-05T03:56:39.462409+00:00"},{"alias_kind":"pith_short_16","alias_value":"BHAUKIF434T4PD6U","created_at":"2026-07-05T03:56:39.462409+00:00"},{"alias_kind":"pith_short_8","alias_value":"BHAUKIF4","created_at":"2026-07-05T03:56:39.462409+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24161","citing_title":"Dual-Branch Cross-Projection Debiasing through Diffusion-based Disentanglement","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03148","citing_title":"$A^2$: Smaller Self-Supervised ViTs Localize Better than Larger Ones","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03148","citing_title":"$A^2$: Smaller Self-Supervised ViTs Localize Better than Larger Ones","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13326","citing_title":"Right Regions, Wrong Labels: Semantic Label Flips in Segmentation under Correlation Shift","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BHAUKIF434T4PD6UWSKNTDIOD5","json":"https://pith.science/pith/BHAUKIF434T4PD6UWSKNTDIOD5.json","graph_json":"https://pith.science/api/pith-number/BHAUKIF434T4PD6UWSKNTDIOD5/graph.json","events_json":"https://pith.science/api/pith-number/BHAUKIF434T4PD6UWSKNTDIOD5/events.json","paper":"https://pith.science/paper/BHAUKIF4"},"agent_actions":{"view_html":"https://pith.science/pith/BHAUKIF434T4PD6UWSKNTDIOD5","download_json":"https://pith.science/pith/BHAUKIF434T4PD6UWSKNTDIOD5.json","view_paper":"https://pith.science/paper/BHAUKIF4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.06523&json=true","fetch_graph":"https://pith.science/api/pith-number/BHAUKIF434T4PD6UWSKNTDIOD5/graph.json","fetch_events":"https://pith.science/api/pith-number/BHAUKIF434T4PD6UWSKNTDIOD5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BHAUKIF434T4PD6UWSKNTDIOD5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BHAUKIF434T4PD6UWSKNTDIOD5/action/storage_attestation","attest_author":"https://pith.science/pith/BHAUKIF434T4PD6UWSKNTDIOD5/action/author_attestation","sign_citation":"https://pith.science/pith/BHAUKIF434T4PD6UWSKNTDIOD5/action/citation_signature","submit_replication":"https://pith.science/pith/BHAUKIF434T4PD6UWSKNTDIOD5/action/replication_record"}},"created_at":"2026-07-05T03:56:39.462409+00:00","updated_at":"2026-07-05T03:56:39.462409+00:00"}