{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:GOM3SWHPECRDFDM7OV25VCOZQX","short_pith_number":"pith:GOM3SWHP","schema_version":"1.0","canonical_sha256":"3399b958ef20a2328d9f7575da89d985cc27f45be491a02f6b7733516a458770","source":{"kind":"arxiv","id":"2203.10163","version":1},"attestation_state":"computed","paper":{"title":"A Closer Look at Knowledge Distillation with Features, Logits, and Gradients","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Hongxia Jin, James Smith, Yen-Chang Hsu, Yilin Shen, Zsolt Kira","submitted_at":"2022-03-18T21:26:55Z","abstract_excerpt":"Knowledge distillation (KD) is a substantial strategy for transferring learned knowledge from one neural network model to another. A vast number of methods have been developed for this strategy. While most method designs a more efficient way to facilitate knowledge transfer, less attention has been put on comparing the effect of knowledge sources such as features, logits, and gradients. This work provides a new perspective to motivate a set of knowledge distillation strategies by approximating the classical KL-divergence criteria with different knowledge sources, making a systematic comparison"},"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":"2203.10163","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2022-03-18T21:26:55Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"15ea168bcd8a05f3287a24de114bb0bbf7c477fb30868ab6ab7c1992eadf636d","abstract_canon_sha256":"6b3c090366057177ad917abc4c06bb77e1bb41c3e481653870e0e8661533502b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:06:32.198932Z","signature_b64":"rMy/0YJo3aD1/49tyu3q42dr7s3DbHwAcIR/Gx7SQd3pls3z1wIJm6MdJgA88XtdIeJJE8fZnLnm52kcFYVJDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3399b958ef20a2328d9f7575da89d985cc27f45be491a02f6b7733516a458770","last_reissued_at":"2026-07-05T04:06:32.198584Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:06:32.198584Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Closer Look at Knowledge Distillation with Features, Logits, and Gradients","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Hongxia Jin, James Smith, Yen-Chang Hsu, Yilin Shen, Zsolt Kira","submitted_at":"2022-03-18T21:26:55Z","abstract_excerpt":"Knowledge distillation (KD) is a substantial strategy for transferring learned knowledge from one neural network model to another. A vast number of methods have been developed for this strategy. While most method designs a more efficient way to facilitate knowledge transfer, less attention has been put on comparing the effect of knowledge sources such as features, logits, and gradients. This work provides a new perspective to motivate a set of knowledge distillation strategies by approximating the classical KL-divergence criteria with different knowledge sources, making a systematic comparison"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.10163","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/2203.10163/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":"2203.10163","created_at":"2026-07-05T04:06:32.198640+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.10163v1","created_at":"2026-07-05T04:06:32.198640+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.10163","created_at":"2026-07-05T04:06:32.198640+00:00"},{"alias_kind":"pith_short_12","alias_value":"GOM3SWHPECRD","created_at":"2026-07-05T04:06:32.198640+00:00"},{"alias_kind":"pith_short_16","alias_value":"GOM3SWHPECRDFDM7","created_at":"2026-07-05T04:06:32.198640+00:00"},{"alias_kind":"pith_short_8","alias_value":"GOM3SWHP","created_at":"2026-07-05T04:06:32.198640+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2509.03934","citing_title":"SelfAug: Mitigating Catastrophic Forgetting in Retrieval-Augmented Generation via Distribution Self-Alignment","ref_index":29,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GOM3SWHPECRDFDM7OV25VCOZQX","json":"https://pith.science/pith/GOM3SWHPECRDFDM7OV25VCOZQX.json","graph_json":"https://pith.science/api/pith-number/GOM3SWHPECRDFDM7OV25VCOZQX/graph.json","events_json":"https://pith.science/api/pith-number/GOM3SWHPECRDFDM7OV25VCOZQX/events.json","paper":"https://pith.science/paper/GOM3SWHP"},"agent_actions":{"view_html":"https://pith.science/pith/GOM3SWHPECRDFDM7OV25VCOZQX","download_json":"https://pith.science/pith/GOM3SWHPECRDFDM7OV25VCOZQX.json","view_paper":"https://pith.science/paper/GOM3SWHP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.10163&json=true","fetch_graph":"https://pith.science/api/pith-number/GOM3SWHPECRDFDM7OV25VCOZQX/graph.json","fetch_events":"https://pith.science/api/pith-number/GOM3SWHPECRDFDM7OV25VCOZQX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GOM3SWHPECRDFDM7OV25VCOZQX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GOM3SWHPECRDFDM7OV25VCOZQX/action/storage_attestation","attest_author":"https://pith.science/pith/GOM3SWHPECRDFDM7OV25VCOZQX/action/author_attestation","sign_citation":"https://pith.science/pith/GOM3SWHPECRDFDM7OV25VCOZQX/action/citation_signature","submit_replication":"https://pith.science/pith/GOM3SWHPECRDFDM7OV25VCOZQX/action/replication_record"}},"created_at":"2026-07-05T04:06:32.198640+00:00","updated_at":"2026-07-05T04:06:32.198640+00:00"}