{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:HDB5PCWIBFP4GCORQ7R6ZKBM5U","short_pith_number":"pith:HDB5PCWI","schema_version":"1.0","canonical_sha256":"38c3d78ac8095fc309d187e3eca82ced04c0bea6a231913e36c1391f10518488","source":{"kind":"arxiv","id":"2402.04050","version":2},"attestation_state":"computed","paper":{"title":"Connecting the Dots: Collaborative Fine-tuning for Black-Box Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Jian Liang, Ran He, Tieniu Tan, Zhengbo Wang, Zilei Wang","submitted_at":"2024-02-06T14:53:19Z","abstract_excerpt":"With the emergence of pretrained vision-language models (VLMs), considerable efforts have been devoted to fine-tuning them for downstream tasks. Despite the progress made in designing efficient fine-tuning methods, such methods require access to the model's parameters, which can be challenging as model owners often opt to provide their models as a black box to safeguard model ownership. This paper proposes a \\textbf{C}ollabo\\textbf{ra}tive \\textbf{F}ine-\\textbf{T}uning (\\textbf{CraFT}) approach for fine-tuning black-box VLMs to downstream tasks, where one only has access to the input prompts a"},"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":"2402.04050","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-06T14:53:19Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"c584c04fbe428013809b1ab472a76fe5e9df4e6c6eaac0fddc1f1a4eb45085b8","abstract_canon_sha256":"60d77f9e236f6038593d2cae79a5c854edf7bce0b9fbf6baf2bc5e187e14ec5f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:26:18.850195Z","signature_b64":"N3S2b1QjGVI3TBHqNxlYTihrDXuB04gjE4Ax4QJ27p2CDIv7AZeUeD/jBVrBs2jEI4XhmTbF7s3He87sIZ3sCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"38c3d78ac8095fc309d187e3eca82ced04c0bea6a231913e36c1391f10518488","last_reissued_at":"2026-07-05T08:26:18.849709Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:26:18.849709Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Connecting the Dots: Collaborative Fine-tuning for Black-Box Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Jian Liang, Ran He, Tieniu Tan, Zhengbo Wang, Zilei Wang","submitted_at":"2024-02-06T14:53:19Z","abstract_excerpt":"With the emergence of pretrained vision-language models (VLMs), considerable efforts have been devoted to fine-tuning them for downstream tasks. Despite the progress made in designing efficient fine-tuning methods, such methods require access to the model's parameters, which can be challenging as model owners often opt to provide their models as a black box to safeguard model ownership. This paper proposes a \\textbf{C}ollabo\\textbf{ra}tive \\textbf{F}ine-\\textbf{T}uning (\\textbf{CraFT}) approach for fine-tuning black-box VLMs to downstream tasks, where one only has access to the input prompts a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.04050","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/2402.04050/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":"2402.04050","created_at":"2026-07-05T08:26:18.849766+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.04050v2","created_at":"2026-07-05T08:26:18.849766+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.04050","created_at":"2026-07-05T08:26:18.849766+00:00"},{"alias_kind":"pith_short_12","alias_value":"HDB5PCWIBFP4","created_at":"2026-07-05T08:26:18.849766+00:00"},{"alias_kind":"pith_short_16","alias_value":"HDB5PCWIBFP4GCOR","created_at":"2026-07-05T08:26:18.849766+00:00"},{"alias_kind":"pith_short_8","alias_value":"HDB5PCWI","created_at":"2026-07-05T08:26:18.849766+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2412.11375","citing_title":"Text and Image Are Mutually Beneficial: Enhancing Training-Free Few-Shot Classification with CLIP","ref_index":50,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HDB5PCWIBFP4GCORQ7R6ZKBM5U","json":"https://pith.science/pith/HDB5PCWIBFP4GCORQ7R6ZKBM5U.json","graph_json":"https://pith.science/api/pith-number/HDB5PCWIBFP4GCORQ7R6ZKBM5U/graph.json","events_json":"https://pith.science/api/pith-number/HDB5PCWIBFP4GCORQ7R6ZKBM5U/events.json","paper":"https://pith.science/paper/HDB5PCWI"},"agent_actions":{"view_html":"https://pith.science/pith/HDB5PCWIBFP4GCORQ7R6ZKBM5U","download_json":"https://pith.science/pith/HDB5PCWIBFP4GCORQ7R6ZKBM5U.json","view_paper":"https://pith.science/paper/HDB5PCWI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.04050&json=true","fetch_graph":"https://pith.science/api/pith-number/HDB5PCWIBFP4GCORQ7R6ZKBM5U/graph.json","fetch_events":"https://pith.science/api/pith-number/HDB5PCWIBFP4GCORQ7R6ZKBM5U/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HDB5PCWIBFP4GCORQ7R6ZKBM5U/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HDB5PCWIBFP4GCORQ7R6ZKBM5U/action/storage_attestation","attest_author":"https://pith.science/pith/HDB5PCWIBFP4GCORQ7R6ZKBM5U/action/author_attestation","sign_citation":"https://pith.science/pith/HDB5PCWIBFP4GCORQ7R6ZKBM5U/action/citation_signature","submit_replication":"https://pith.science/pith/HDB5PCWIBFP4GCORQ7R6ZKBM5U/action/replication_record"}},"created_at":"2026-07-05T08:26:18.849766+00:00","updated_at":"2026-07-05T08:26:18.849766+00:00"}