{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:3O4VCUK2SKMP2GBME67F7IEQ36","short_pith_number":"pith:3O4VCUK2","schema_version":"1.0","canonical_sha256":"dbb951515a9298fd182c27be5fa090dfb7b30b6b1320f57c86a5be8ddd202e6b","source":{"kind":"arxiv","id":"1909.09656","version":2},"attestation_state":"computed","paper":{"title":"Understanding and Robustifying Differentiable Architecture Search","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Arber Zela, Frank Hutter, Thomas Brox, Thomas Elsken, Tonmoy Saikia, Yassine Marrakchi","submitted_at":"2019-09-20T18:03:06Z","abstract_excerpt":"Differentiable Architecture Search (DARTS) has attracted a lot of attention due to its simplicity and small search costs achieved by a continuous relaxation and an approximation of the resulting bi-level optimization problem. However, DARTS does not work robustly for new problems: we identify a wide range of search spaces for which DARTS yields degenerate architectures with very poor test performance. We study this failure mode and show that, while DARTS successfully minimizes validation loss, the found solutions generalize poorly when they coincide with high validation loss curvature in the a"},"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":"1909.09656","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-09-20T18:03:06Z","cross_cats_sorted":["cs.AI","cs.CV","stat.ML"],"title_canon_sha256":"c654da795c39079b86ebb0fbe25aab1d24fbcb74522eb73efa8b81b7032cbfd8","abstract_canon_sha256":"c690709896fb0b5594e9a5730573c2c02378bb272f1da288bc730f13d06b231d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:36:37.506814Z","signature_b64":"hpJTbBXqc/to7GHRbfinOwB0vYMl3YZ5cx8E5ZZLEtYKCF97BXoN7/HCRLurJFodtbn59l+kRXTZjKHgjhFICw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dbb951515a9298fd182c27be5fa090dfb7b30b6b1320f57c86a5be8ddd202e6b","last_reissued_at":"2026-07-05T00:36:37.506408Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:36:37.506408Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding and Robustifying Differentiable Architecture Search","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Arber Zela, Frank Hutter, Thomas Brox, Thomas Elsken, Tonmoy Saikia, Yassine Marrakchi","submitted_at":"2019-09-20T18:03:06Z","abstract_excerpt":"Differentiable Architecture Search (DARTS) has attracted a lot of attention due to its simplicity and small search costs achieved by a continuous relaxation and an approximation of the resulting bi-level optimization problem. However, DARTS does not work robustly for new problems: we identify a wide range of search spaces for which DARTS yields degenerate architectures with very poor test performance. We study this failure mode and show that, while DARTS successfully minimizes validation loss, the found solutions generalize poorly when they coincide with high validation loss curvature in the a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1909.09656","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/1909.09656/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":"1909.09656","created_at":"2026-07-05T00:36:37.506464+00:00"},{"alias_kind":"arxiv_version","alias_value":"1909.09656v2","created_at":"2026-07-05T00:36:37.506464+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1909.09656","created_at":"2026-07-05T00:36:37.506464+00:00"},{"alias_kind":"pith_short_12","alias_value":"3O4VCUK2SKMP","created_at":"2026-07-05T00:36:37.506464+00:00"},{"alias_kind":"pith_short_16","alias_value":"3O4VCUK2SKMP2GBM","created_at":"2026-07-05T00:36:37.506464+00:00"},{"alias_kind":"pith_short_8","alias_value":"3O4VCUK2","created_at":"2026-07-05T00:36:37.506464+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.29582","citing_title":"Bilevel Optimization for Neural Architecture Search","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29518","citing_title":"Harvesting AI Computation at the Edge via Generic Approximation","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11948","citing_title":"Active Imitation Learning for Thermal- and Kernel-Aware LFM Inference on 3D S-NUCA Many-Cores","ref_index":53,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3O4VCUK2SKMP2GBME67F7IEQ36","json":"https://pith.science/pith/3O4VCUK2SKMP2GBME67F7IEQ36.json","graph_json":"https://pith.science/api/pith-number/3O4VCUK2SKMP2GBME67F7IEQ36/graph.json","events_json":"https://pith.science/api/pith-number/3O4VCUK2SKMP2GBME67F7IEQ36/events.json","paper":"https://pith.science/paper/3O4VCUK2"},"agent_actions":{"view_html":"https://pith.science/pith/3O4VCUK2SKMP2GBME67F7IEQ36","download_json":"https://pith.science/pith/3O4VCUK2SKMP2GBME67F7IEQ36.json","view_paper":"https://pith.science/paper/3O4VCUK2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1909.09656&json=true","fetch_graph":"https://pith.science/api/pith-number/3O4VCUK2SKMP2GBME67F7IEQ36/graph.json","fetch_events":"https://pith.science/api/pith-number/3O4VCUK2SKMP2GBME67F7IEQ36/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3O4VCUK2SKMP2GBME67F7IEQ36/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3O4VCUK2SKMP2GBME67F7IEQ36/action/storage_attestation","attest_author":"https://pith.science/pith/3O4VCUK2SKMP2GBME67F7IEQ36/action/author_attestation","sign_citation":"https://pith.science/pith/3O4VCUK2SKMP2GBME67F7IEQ36/action/citation_signature","submit_replication":"https://pith.science/pith/3O4VCUK2SKMP2GBME67F7IEQ36/action/replication_record"}},"created_at":"2026-07-05T00:36:37.506464+00:00","updated_at":"2026-07-05T00:36:37.506464+00:00"}