{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:XEX424CZU4RWA5P47WC55EYX2L","short_pith_number":"pith:XEX424CZ","schema_version":"1.0","canonical_sha256":"b92fcd7059a7236075fcfd85de9317d2e561f587059bcc65c44c86083c7f3d04","source":{"kind":"arxiv","id":"2110.06893","version":3},"attestation_state":"computed","paper":{"title":"Newer is not always better: Rethinking transferability metrics, their peculiarities, stability and performance","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Natalia Ponomareva, Rahul Mazumder, Shibal Ibrahim","submitted_at":"2021-10-13T17:24:12Z","abstract_excerpt":"Fine-tuning of large pre-trained image and language models on small customized datasets has become increasingly popular for improved prediction and efficient use of limited resources. Fine-tuning requires identification of best models to transfer-learn from and quantifying transferability prevents expensive re-training on all of the candidate models/tasks pairs. In this paper, we show that the statistical problems with covariance estimation drive the poor performance of H-score -- a common baseline for newer metrics -- and propose shrinkage-based estimator. This results in up to 80% absolute g"},"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":"2110.06893","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-10-13T17:24:12Z","cross_cats_sorted":[],"title_canon_sha256":"1f34877a182c7505295ada3a3c714683aa0ce186e5f5b0803e69fc9a3681c133","abstract_canon_sha256":"163bf468d297d2bd5b67b0d1bbc6ed0a9c5624d94a2f874d80b502033145b5bb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:14:02.973823Z","signature_b64":"ecbLIQVW5LHQ9+Gka5ALl89VsYOyBLBM6rkYyD4MBKfl4gkACH8AK8kGoV83Ingcse848BqGdGcGzT3pNUJkDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b92fcd7059a7236075fcfd85de9317d2e561f587059bcc65c44c86083c7f3d04","last_reissued_at":"2026-07-05T06:14:02.973478Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:14:02.973478Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Newer is not always better: Rethinking transferability metrics, their peculiarities, stability and performance","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Natalia Ponomareva, Rahul Mazumder, Shibal Ibrahim","submitted_at":"2021-10-13T17:24:12Z","abstract_excerpt":"Fine-tuning of large pre-trained image and language models on small customized datasets has become increasingly popular for improved prediction and efficient use of limited resources. Fine-tuning requires identification of best models to transfer-learn from and quantifying transferability prevents expensive re-training on all of the candidate models/tasks pairs. In this paper, we show that the statistical problems with covariance estimation drive the poor performance of H-score -- a common baseline for newer metrics -- and propose shrinkage-based estimator. This results in up to 80% absolute g"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2110.06893","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/2110.06893/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":"2110.06893","created_at":"2026-07-05T06:14:02.973535+00:00"},{"alias_kind":"arxiv_version","alias_value":"2110.06893v3","created_at":"2026-07-05T06:14:02.973535+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2110.06893","created_at":"2026-07-05T06:14:02.973535+00:00"},{"alias_kind":"pith_short_12","alias_value":"XEX424CZU4RW","created_at":"2026-07-05T06:14:02.973535+00:00"},{"alias_kind":"pith_short_16","alias_value":"XEX424CZU4RWA5P4","created_at":"2026-07-05T06:14:02.973535+00:00"},{"alias_kind":"pith_short_8","alias_value":"XEX424CZ","created_at":"2026-07-05T06:14:02.973535+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2503.00450","citing_title":"Unsupervised Source-Free Ranking of Biomedical Segmentation Models Under Distribution Shift","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XEX424CZU4RWA5P47WC55EYX2L","json":"https://pith.science/pith/XEX424CZU4RWA5P47WC55EYX2L.json","graph_json":"https://pith.science/api/pith-number/XEX424CZU4RWA5P47WC55EYX2L/graph.json","events_json":"https://pith.science/api/pith-number/XEX424CZU4RWA5P47WC55EYX2L/events.json","paper":"https://pith.science/paper/XEX424CZ"},"agent_actions":{"view_html":"https://pith.science/pith/XEX424CZU4RWA5P47WC55EYX2L","download_json":"https://pith.science/pith/XEX424CZU4RWA5P47WC55EYX2L.json","view_paper":"https://pith.science/paper/XEX424CZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2110.06893&json=true","fetch_graph":"https://pith.science/api/pith-number/XEX424CZU4RWA5P47WC55EYX2L/graph.json","fetch_events":"https://pith.science/api/pith-number/XEX424CZU4RWA5P47WC55EYX2L/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XEX424CZU4RWA5P47WC55EYX2L/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XEX424CZU4RWA5P47WC55EYX2L/action/storage_attestation","attest_author":"https://pith.science/pith/XEX424CZU4RWA5P47WC55EYX2L/action/author_attestation","sign_citation":"https://pith.science/pith/XEX424CZU4RWA5P47WC55EYX2L/action/citation_signature","submit_replication":"https://pith.science/pith/XEX424CZU4RWA5P47WC55EYX2L/action/replication_record"}},"created_at":"2026-07-05T06:14:02.973535+00:00","updated_at":"2026-07-05T06:14:02.973535+00:00"}