{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:K6UK4PT5TRSWOTL3BB3YY476SJ","short_pith_number":"pith:K6UK4PT5","schema_version":"1.0","canonical_sha256":"57a8ae3e7d9c65674d7b08778c73fe92574976910e878703cb7367396ae88269","source":{"kind":"arxiv","id":"2412.07226","version":1},"attestation_state":"computed","paper":{"title":"Attention Head Purification: A New Perspective to Harness CLIP for Domain Generalization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Guoliang Kang, Yingfan Wang","submitted_at":"2024-12-10T06:27:48Z","abstract_excerpt":"Domain Generalization (DG) aims to learn a model from multiple source domains to achieve satisfactory performance on unseen target domains. Recent works introduce CLIP to DG tasks due to its superior image-text alignment and zeros-shot performance. Previous methods either utilize full fine-tuning or prompt-learning paradigms to harness CLIP for DG tasks. Those works focus on avoiding catastrophic forgetting of the original knowledge encoded in CLIP but ignore that the knowledge encoded in CLIP in nature may contain domain-specific cues that constrain its domain generalization performance. In t"},"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":"2412.07226","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-12-10T06:27:48Z","cross_cats_sorted":[],"title_canon_sha256":"57bc4468012406da2879a7066f2ade0a11bb0aec5b3d8cfe0173133e34bc0f49","abstract_canon_sha256":"cef2f5fa131a8be366ad87d00c880e5958aabd49f5611584bfd63695e613b5f5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:47:06.309574Z","signature_b64":"emiC2WH4rTCwwK+wQ8XQ3H/5O6xH7DwSrMFsPkwwAq3m2fOAoXtdLJTJGmS+b3NB+XBUarGKo+AOBqjaCm2NAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"57a8ae3e7d9c65674d7b08778c73fe92574976910e878703cb7367396ae88269","last_reissued_at":"2026-07-05T09:47:06.308985Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:47:06.308985Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Attention Head Purification: A New Perspective to Harness CLIP for Domain Generalization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Guoliang Kang, Yingfan Wang","submitted_at":"2024-12-10T06:27:48Z","abstract_excerpt":"Domain Generalization (DG) aims to learn a model from multiple source domains to achieve satisfactory performance on unseen target domains. Recent works introduce CLIP to DG tasks due to its superior image-text alignment and zeros-shot performance. Previous methods either utilize full fine-tuning or prompt-learning paradigms to harness CLIP for DG tasks. Those works focus on avoiding catastrophic forgetting of the original knowledge encoded in CLIP but ignore that the knowledge encoded in CLIP in nature may contain domain-specific cues that constrain its domain generalization performance. In t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.07226","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/2412.07226/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":"2412.07226","created_at":"2026-07-05T09:47:06.309049+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.07226v1","created_at":"2026-07-05T09:47:06.309049+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.07226","created_at":"2026-07-05T09:47:06.309049+00:00"},{"alias_kind":"pith_short_12","alias_value":"K6UK4PT5TRSW","created_at":"2026-07-05T09:47:06.309049+00:00"},{"alias_kind":"pith_short_16","alias_value":"K6UK4PT5TRSWOTL3","created_at":"2026-07-05T09:47:06.309049+00:00"},{"alias_kind":"pith_short_8","alias_value":"K6UK4PT5","created_at":"2026-07-05T09:47:06.309049+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.06775","citing_title":"Enhancing Performance of Explainable AI Models with Constrained Concept Refinement","ref_index":63,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/K6UK4PT5TRSWOTL3BB3YY476SJ","json":"https://pith.science/pith/K6UK4PT5TRSWOTL3BB3YY476SJ.json","graph_json":"https://pith.science/api/pith-number/K6UK4PT5TRSWOTL3BB3YY476SJ/graph.json","events_json":"https://pith.science/api/pith-number/K6UK4PT5TRSWOTL3BB3YY476SJ/events.json","paper":"https://pith.science/paper/K6UK4PT5"},"agent_actions":{"view_html":"https://pith.science/pith/K6UK4PT5TRSWOTL3BB3YY476SJ","download_json":"https://pith.science/pith/K6UK4PT5TRSWOTL3BB3YY476SJ.json","view_paper":"https://pith.science/paper/K6UK4PT5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.07226&json=true","fetch_graph":"https://pith.science/api/pith-number/K6UK4PT5TRSWOTL3BB3YY476SJ/graph.json","fetch_events":"https://pith.science/api/pith-number/K6UK4PT5TRSWOTL3BB3YY476SJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/K6UK4PT5TRSWOTL3BB3YY476SJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/K6UK4PT5TRSWOTL3BB3YY476SJ/action/storage_attestation","attest_author":"https://pith.science/pith/K6UK4PT5TRSWOTL3BB3YY476SJ/action/author_attestation","sign_citation":"https://pith.science/pith/K6UK4PT5TRSWOTL3BB3YY476SJ/action/citation_signature","submit_replication":"https://pith.science/pith/K6UK4PT5TRSWOTL3BB3YY476SJ/action/replication_record"}},"created_at":"2026-07-05T09:47:06.309049+00:00","updated_at":"2026-07-05T09:47:06.309049+00:00"}