{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2020:FQLTFPBHDFXLOGQOHTO6NHGDKZ","short_pith_number":"pith:FQLTFPBH","canonical_record":{"source":{"id":"2010.06866","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-10-14T07:59:00Z","cross_cats_sorted":["cs.CV","stat.ML"],"title_canon_sha256":"47a6560327c9cd2abfb61e36d50af4c5ec24bf6c58820b4fb07d0299df265318","abstract_canon_sha256":"c3ddff02ececbdf2f7971b9f8e4624e559e5d79b787c4de9c55dd8246d1d440e"},"schema_version":"1.0"},"canonical_sha256":"2c1732bc27196eb71a0e3cdde69cc356641255b2f0b61885f2de21d976579a02","source":{"kind":"arxiv","id":"2010.06866","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2010.06866","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"arxiv_version","alias_value":"2010.06866v2","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2010.06866","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"pith_short_12","alias_value":"FQLTFPBHDFXL","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"pith_short_16","alias_value":"FQLTFPBHDFXLOGQO","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"pith_short_8","alias_value":"FQLTFPBH","created_at":"2026-07-05T01:43:59Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2020:FQLTFPBHDFXLOGQOHTO6NHGDKZ","target":"record","payload":{"canonical_record":{"source":{"id":"2010.06866","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-10-14T07:59:00Z","cross_cats_sorted":["cs.CV","stat.ML"],"title_canon_sha256":"47a6560327c9cd2abfb61e36d50af4c5ec24bf6c58820b4fb07d0299df265318","abstract_canon_sha256":"c3ddff02ececbdf2f7971b9f8e4624e559e5d79b787c4de9c55dd8246d1d440e"},"schema_version":"1.0"},"canonical_sha256":"2c1732bc27196eb71a0e3cdde69cc356641255b2f0b61885f2de21d976579a02","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:43:59.743682Z","signature_b64":"DghdzYgGqq/x/CtvWxpTt2WBKCGVDxGnR11pYYC12egSYQMrLonhYIsYx5toiAO9efDi5WTRqbXLZSiAcyZ0Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2c1732bc27196eb71a0e3cdde69cc356641255b2f0b61885f2de21d976579a02","last_reissued_at":"2026-07-05T01:43:59.743200Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:43:59.743200Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2010.06866","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T01:43:59Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ZOCcQvsXJjEPawA7Br/Ab8TIdPK23tJhx751G1XFEUlRC4ps/0VyzMMdBOv3ud/kfM36HOxiHX5Rq8DYE6BsCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T07:11:17.295798Z"},"content_sha256":"76589162f9c450e2335654ec22f3e0a8436278fbcac408fa895e117badeb8abb","schema_version":"1.0","event_id":"sha256:76589162f9c450e2335654ec22f3e0a8436278fbcac408fa895e117badeb8abb"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2020:FQLTFPBHDFXLOGQOHTO6NHGDKZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Deep Ensembles for Low-Data Transfer Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andr\\'e Susano Pinto, Basil Mustafa, Carlos Riquelme, Daniel Keysers, Joan Puigcerver, Neil Houlsby","submitted_at":"2020-10-14T07:59:00Z","abstract_excerpt":"In the low-data regime, it is difficult to train good supervised models from scratch. Instead practitioners turn to pre-trained models, leveraging transfer learning. Ensembling is an empirically and theoretically appealing way to construct powerful predictive models, but the predominant approach of training multiple deep networks with different random initialisations collides with the need for transfer via pre-trained weights. In this work, we study different ways of creating ensembles from pre-trained models. We show that the nature of pre-training itself is a performant source of diversity, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2010.06866","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/2010.06866/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T01:43:59Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"zyvknvrVgk+EogcyGYyghbmbqRZKULO3Y9RzSj01PuPjdFtsrUzoXncvQj1QP5Ta1nbjE4BAKf2c2LY5lS0qAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T07:11:17.296310Z"},"content_sha256":"7f1418c61dc20343f54946528b8f10a770644be1f69a00a26b95de06522d5b0f","schema_version":"1.0","event_id":"sha256:7f1418c61dc20343f54946528b8f10a770644be1f69a00a26b95de06522d5b0f"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/FQLTFPBHDFXLOGQOHTO6NHGDKZ/bundle.json","state_url":"https://pith.science/pith/FQLTFPBHDFXLOGQOHTO6NHGDKZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/FQLTFPBHDFXLOGQOHTO6NHGDKZ/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-08T07:11:17Z","links":{"resolver":"https://pith.science/pith/FQLTFPBHDFXLOGQOHTO6NHGDKZ","bundle":"https://pith.science/pith/FQLTFPBHDFXLOGQOHTO6NHGDKZ/bundle.json","state":"https://pith.science/pith/FQLTFPBHDFXLOGQOHTO6NHGDKZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/FQLTFPBHDFXLOGQOHTO6NHGDKZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2020:FQLTFPBHDFXLOGQOHTO6NHGDKZ","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"c3ddff02ececbdf2f7971b9f8e4624e559e5d79b787c4de9c55dd8246d1d440e","cross_cats_sorted":["cs.CV","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-10-14T07:59:00Z","title_canon_sha256":"47a6560327c9cd2abfb61e36d50af4c5ec24bf6c58820b4fb07d0299df265318"},"schema_version":"1.0","source":{"id":"2010.06866","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2010.06866","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"arxiv_version","alias_value":"2010.06866v2","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2010.06866","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"pith_short_12","alias_value":"FQLTFPBHDFXL","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"pith_short_16","alias_value":"FQLTFPBHDFXLOGQO","created_at":"2026-07-05T01:43:59Z"},{"alias_kind":"pith_short_8","alias_value":"FQLTFPBH","created_at":"2026-07-05T01:43:59Z"}],"graph_snapshots":[{"event_id":"sha256:7f1418c61dc20343f54946528b8f10a770644be1f69a00a26b95de06522d5b0f","target":"graph","created_at":"2026-07-05T01:43:59Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2010.06866/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"In the low-data regime, it is difficult to train good supervised models from scratch. Instead practitioners turn to pre-trained models, leveraging transfer learning. Ensembling is an empirically and theoretically appealing way to construct powerful predictive models, but the predominant approach of training multiple deep networks with different random initialisations collides with the need for transfer via pre-trained weights. In this work, we study different ways of creating ensembles from pre-trained models. We show that the nature of pre-training itself is a performant source of diversity, ","authors_text":"Andr\\'e Susano Pinto, Basil Mustafa, Carlos Riquelme, Daniel Keysers, Joan Puigcerver, Neil Houlsby","cross_cats":["cs.CV","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-10-14T07:59:00Z","title":"Deep Ensembles for Low-Data Transfer Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2010.06866","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:76589162f9c450e2335654ec22f3e0a8436278fbcac408fa895e117badeb8abb","target":"record","created_at":"2026-07-05T01:43:59Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"c3ddff02ececbdf2f7971b9f8e4624e559e5d79b787c4de9c55dd8246d1d440e","cross_cats_sorted":["cs.CV","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-10-14T07:59:00Z","title_canon_sha256":"47a6560327c9cd2abfb61e36d50af4c5ec24bf6c58820b4fb07d0299df265318"},"schema_version":"1.0","source":{"id":"2010.06866","kind":"arxiv","version":2}},"canonical_sha256":"2c1732bc27196eb71a0e3cdde69cc356641255b2f0b61885f2de21d976579a02","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2c1732bc27196eb71a0e3cdde69cc356641255b2f0b61885f2de21d976579a02","first_computed_at":"2026-07-05T01:43:59.743200Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T01:43:59.743200Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"DghdzYgGqq/x/CtvWxpTt2WBKCGVDxGnR11pYYC12egSYQMrLonhYIsYx5toiAO9efDi5WTRqbXLZSiAcyZ0Ag==","signature_status":"signed_v1","signed_at":"2026-07-05T01:43:59.743682Z","signed_message":"canonical_sha256_bytes"},"source_id":"2010.06866","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:76589162f9c450e2335654ec22f3e0a8436278fbcac408fa895e117badeb8abb","sha256:7f1418c61dc20343f54946528b8f10a770644be1f69a00a26b95de06522d5b0f"],"state_sha256":"419a8ae7aef00a1c1fc5786c08df3f9d237512278a0e79f0d555f60437b0a6fd"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"JEp1pacMmBrpI80L6WzqWkfmcoAZVOY7PYAYVMCV9nA2IbAZR7r+2hEhSydf9Km9+D4tvp9dh1u/jd3CNy3ICg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T07:11:17.301281Z","bundle_sha256":"ae1b8a2a2f290b4793bd2125d230c82a925601ce5c12ae2d85b413959a8aeb4e"}}