{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:W2Z6UKKCQ2CVZ727OPQ7YLRZ5W","short_pith_number":"pith:W2Z6UKKC","schema_version":"1.0","canonical_sha256":"b6b3ea294286855cff5f73e1fc2e39ed9872e5ce718d0e2659997434361ed30d","source":{"kind":"arxiv","id":"2002.05715","version":3},"attestation_state":"computed","paper":{"title":"Self-Distillation Amplifies Regularization in Hilbert Space","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Hossein Mobahi, Mehrdad Farajtabar, Peter L. Bartlett","submitted_at":"2020-02-13T18:56:06Z","abstract_excerpt":"Knowledge distillation introduced in the deep learning context is a method to transfer knowledge from one architecture to another. In particular, when the architectures are identical, this is called self-distillation. The idea is to feed in predictions of the trained model as new target values for retraining (and iterate this loop possibly a few times). It has been empirically observed that the self-distilled model often achieves higher accuracy on held out data. Why this happens, however, has been a mystery: the self-distillation dynamics does not receive any new information about the task an"},"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":"2002.05715","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-02-13T18:56:06Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"800de8f48abcba30c2427972424d251381bd88b0ca6ac4335f23320a5e03925c","abstract_canon_sha256":"65a03a2d4284546edd00323e8e1cb98d4a6e50ab20402aa566dd313f6ed8c7f1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:46:00.124872Z","signature_b64":"KHe5XC4VezdJkh/Si3tYZOdPBPPgz+OuCsg6mkFVJocUmOEG7eEjENCtnscW+9s6zMS1+Nmx+WM7Jpl/0Nk+Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b6b3ea294286855cff5f73e1fc2e39ed9872e5ce718d0e2659997434361ed30d","last_reissued_at":"2026-07-05T01:46:00.124408Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:46:00.124408Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Distillation Amplifies Regularization in Hilbert Space","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Hossein Mobahi, Mehrdad Farajtabar, Peter L. Bartlett","submitted_at":"2020-02-13T18:56:06Z","abstract_excerpt":"Knowledge distillation introduced in the deep learning context is a method to transfer knowledge from one architecture to another. In particular, when the architectures are identical, this is called self-distillation. The idea is to feed in predictions of the trained model as new target values for retraining (and iterate this loop possibly a few times). It has been empirically observed that the self-distilled model often achieves higher accuracy on held out data. Why this happens, however, has been a mystery: the self-distillation dynamics does not receive any new information about the task an"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2002.05715","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/2002.05715/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":"2002.05715","created_at":"2026-07-05T01:46:00.124465+00:00"},{"alias_kind":"arxiv_version","alias_value":"2002.05715v3","created_at":"2026-07-05T01:46:00.124465+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2002.05715","created_at":"2026-07-05T01:46:00.124465+00:00"},{"alias_kind":"pith_short_12","alias_value":"W2Z6UKKCQ2CV","created_at":"2026-07-05T01:46:00.124465+00:00"},{"alias_kind":"pith_short_16","alias_value":"W2Z6UKKCQ2CVZ727","created_at":"2026-07-05T01:46:00.124465+00:00"},{"alias_kind":"pith_short_8","alias_value":"W2Z6UKKC","created_at":"2026-07-05T01:46:00.124465+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.03938","citing_title":"q0: Primitives for Hyper-Epoch Pretraining","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2602.12687","citing_title":"Trust the uncertain teacher: distilling dark knowledge via calibrated uncertainty","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W","json":"https://pith.science/pith/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W.json","graph_json":"https://pith.science/api/pith-number/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W/graph.json","events_json":"https://pith.science/api/pith-number/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W/events.json","paper":"https://pith.science/paper/W2Z6UKKC"},"agent_actions":{"view_html":"https://pith.science/pith/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W","download_json":"https://pith.science/pith/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W.json","view_paper":"https://pith.science/paper/W2Z6UKKC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2002.05715&json=true","fetch_graph":"https://pith.science/api/pith-number/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W/graph.json","fetch_events":"https://pith.science/api/pith-number/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W/action/storage_attestation","attest_author":"https://pith.science/pith/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W/action/author_attestation","sign_citation":"https://pith.science/pith/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W/action/citation_signature","submit_replication":"https://pith.science/pith/W2Z6UKKCQ2CVZ727OPQ7YLRZ5W/action/replication_record"}},"created_at":"2026-07-05T01:46:00.124465+00:00","updated_at":"2026-07-05T01:46:00.124465+00:00"}