{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:QGK4FA3E73ODQORY3NJH2KMUYN","short_pith_number":"pith:QGK4FA3E","schema_version":"1.0","canonical_sha256":"8195c28364fedc383a38db527d2994c357eeb69697a271b603dafe6c385c472e","source":{"kind":"arxiv","id":"2310.15308","version":4},"attestation_state":"computed","paper":{"title":"SAM-CLIP: Merging Vision Foundation Models towards Semantic and Spatial Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Fartash Faghri, Hadi Pouransari, Haoxiang Wang, Mehrdad Farajtabar, Mohammad Rastegari, Oncel Tuzel, Pavan Kumar Anasosalu Vasu, Raviteja Vemulapalli, Sachin Mehta","submitted_at":"2023-10-23T19:21:57Z","abstract_excerpt":"The landscape of publicly available vision foundation models (VFMs), such as CLIP and Segment Anything Model (SAM), is expanding rapidly. VFMs are endowed with distinct capabilities stemming from their pre-training objectives. For instance, CLIP excels in semantic understanding, while SAM specializes in spatial understanding for segmentation. In this work, we introduce a simple recipe to efficiently merge VFMs into a unified model that absorbs their expertise. Our method integrates techniques of multi-task learning, continual learning, and distillation. Further, it demands significantly less c"},"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":"2310.15308","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-10-23T19:21:57Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"5f203d666209809cda12d06b3172103b63dd716a405db02a1783b976eb340f82","abstract_canon_sha256":"64b851c31bf2960e88a5266cef2c0a8b9d7d8516604405da4052a1b925b93ff0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:29:45.080149Z","signature_b64":"i/DiOGOYJOFasU1atFrZpKft9Fn2QigYRBcXIFohzVB+QtMsu7Bame20b7dgVxUi9T7vzMcNj6Jv+QsvwBMLCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8195c28364fedc383a38db527d2994c357eeb69697a271b603dafe6c385c472e","last_reissued_at":"2026-07-05T08:29:45.079698Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:29:45.079698Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SAM-CLIP: Merging Vision Foundation Models towards Semantic and Spatial Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Fartash Faghri, Hadi Pouransari, Haoxiang Wang, Mehrdad Farajtabar, Mohammad Rastegari, Oncel Tuzel, Pavan Kumar Anasosalu Vasu, Raviteja Vemulapalli, Sachin Mehta","submitted_at":"2023-10-23T19:21:57Z","abstract_excerpt":"The landscape of publicly available vision foundation models (VFMs), such as CLIP and Segment Anything Model (SAM), is expanding rapidly. VFMs are endowed with distinct capabilities stemming from their pre-training objectives. For instance, CLIP excels in semantic understanding, while SAM specializes in spatial understanding for segmentation. In this work, we introduce a simple recipe to efficiently merge VFMs into a unified model that absorbs their expertise. Our method integrates techniques of multi-task learning, continual learning, and distillation. Further, it demands significantly less c"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.15308","kind":"arxiv","version":4},"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/2310.15308/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":"2310.15308","created_at":"2026-07-05T08:29:45.079757+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.15308v4","created_at":"2026-07-05T08:29:45.079757+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.15308","created_at":"2026-07-05T08:29:45.079757+00:00"},{"alias_kind":"pith_short_12","alias_value":"QGK4FA3E73OD","created_at":"2026-07-05T08:29:45.079757+00:00"},{"alias_kind":"pith_short_16","alias_value":"QGK4FA3E73ODQORY","created_at":"2026-07-05T08:29:45.079757+00:00"},{"alias_kind":"pith_short_8","alias_value":"QGK4FA3E","created_at":"2026-07-05T08:29:45.079757+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.19410","citing_title":"Vision Harnessing Agent for Open Ad-hoc Segmentation","ref_index":61,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QGK4FA3E73ODQORY3NJH2KMUYN","json":"https://pith.science/pith/QGK4FA3E73ODQORY3NJH2KMUYN.json","graph_json":"https://pith.science/api/pith-number/QGK4FA3E73ODQORY3NJH2KMUYN/graph.json","events_json":"https://pith.science/api/pith-number/QGK4FA3E73ODQORY3NJH2KMUYN/events.json","paper":"https://pith.science/paper/QGK4FA3E"},"agent_actions":{"view_html":"https://pith.science/pith/QGK4FA3E73ODQORY3NJH2KMUYN","download_json":"https://pith.science/pith/QGK4FA3E73ODQORY3NJH2KMUYN.json","view_paper":"https://pith.science/paper/QGK4FA3E","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.15308&json=true","fetch_graph":"https://pith.science/api/pith-number/QGK4FA3E73ODQORY3NJH2KMUYN/graph.json","fetch_events":"https://pith.science/api/pith-number/QGK4FA3E73ODQORY3NJH2KMUYN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QGK4FA3E73ODQORY3NJH2KMUYN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QGK4FA3E73ODQORY3NJH2KMUYN/action/storage_attestation","attest_author":"https://pith.science/pith/QGK4FA3E73ODQORY3NJH2KMUYN/action/author_attestation","sign_citation":"https://pith.science/pith/QGK4FA3E73ODQORY3NJH2KMUYN/action/citation_signature","submit_replication":"https://pith.science/pith/QGK4FA3E73ODQORY3NJH2KMUYN/action/replication_record"}},"created_at":"2026-07-05T08:29:45.079757+00:00","updated_at":"2026-07-05T08:29:45.079757+00:00"}