{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4EXUYT2K3TMUXCXJ3C5H637366","short_pith_number":"pith:4EXUYT2K","schema_version":"1.0","canonical_sha256":"e12f4c4f4adcd94b8ae9d8ba7f6ffbf79a0af165761fb217fb312146eb961216","source":{"kind":"arxiv","id":"2402.16832","version":2},"attestation_state":"computed","paper":{"title":"Cross-Modal Projection in Multimodal LLMs Doesn't Really Project Visual Attributes to Textual Space","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.CL","authors_text":"Gaurav Verma, Jamelle Watson-Daniels, Kartik Sharma, Minje Choi, Sejoon Oh, Srijan Kumar","submitted_at":"2024-02-26T18:56:48Z","abstract_excerpt":"Multimodal large language models (MLLMs) like LLaVA and GPT-4(V) enable general-purpose conversations about images with the language modality. As off-the-shelf MLLMs may have limited capabilities on images from domains like dermatology and agriculture, they must be fine-tuned to unlock domain-specific applications. The prevalent architecture of current open-source MLLMs comprises two major modules: an image-language (cross-modal) projection network and a large language model. It is desirable to understand the roles of these two modules in modeling domain-specific visual attributes to inform th"},"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":"2402.16832","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-26T18:56:48Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"e24beb95e072cd52daaf39314ff1dac25854e2c501d1dc6176c356165683e700","abstract_canon_sha256":"6a6a88d88fbaf235f411fec691204346f47900e6f49e1cdc25bf5c6c37c21c32"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:46:22.011611Z","signature_b64":"Dzo/wSKGHJOAZSiXg4Y5CWNUl0PEn/CCoHEqKD1D9uVXONf0v6naydHoHFcDmNxWPV2Mx7zADxndC1N9OmLYAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e12f4c4f4adcd94b8ae9d8ba7f6ffbf79a0af165761fb217fb312146eb961216","last_reissued_at":"2026-07-05T08:46:22.011176Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:46:22.011176Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Cross-Modal Projection in Multimodal LLMs Doesn't Really Project Visual Attributes to Textual Space","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.CL","authors_text":"Gaurav Verma, Jamelle Watson-Daniels, Kartik Sharma, Minje Choi, Sejoon Oh, Srijan Kumar","submitted_at":"2024-02-26T18:56:48Z","abstract_excerpt":"Multimodal large language models (MLLMs) like LLaVA and GPT-4(V) enable general-purpose conversations about images with the language modality. As off-the-shelf MLLMs may have limited capabilities on images from domains like dermatology and agriculture, they must be fine-tuned to unlock domain-specific applications. The prevalent architecture of current open-source MLLMs comprises two major modules: an image-language (cross-modal) projection network and a large language model. It is desirable to understand the roles of these two modules in modeling domain-specific visual attributes to inform th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.16832","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/2402.16832/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":"2402.16832","created_at":"2026-07-05T08:46:22.011237+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.16832v2","created_at":"2026-07-05T08:46:22.011237+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.16832","created_at":"2026-07-05T08:46:22.011237+00:00"},{"alias_kind":"pith_short_12","alias_value":"4EXUYT2K3TMU","created_at":"2026-07-05T08:46:22.011237+00:00"},{"alias_kind":"pith_short_16","alias_value":"4EXUYT2K3TMUXCXJ","created_at":"2026-07-05T08:46:22.011237+00:00"},{"alias_kind":"pith_short_8","alias_value":"4EXUYT2K","created_at":"2026-07-05T08:46:22.011237+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.22989","citing_title":"CheXmix: Unified Generative Pretraining for Vision Language Models in Medical Imaging","ref_index":41,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4EXUYT2K3TMUXCXJ3C5H637366","json":"https://pith.science/pith/4EXUYT2K3TMUXCXJ3C5H637366.json","graph_json":"https://pith.science/api/pith-number/4EXUYT2K3TMUXCXJ3C5H637366/graph.json","events_json":"https://pith.science/api/pith-number/4EXUYT2K3TMUXCXJ3C5H637366/events.json","paper":"https://pith.science/paper/4EXUYT2K"},"agent_actions":{"view_html":"https://pith.science/pith/4EXUYT2K3TMUXCXJ3C5H637366","download_json":"https://pith.science/pith/4EXUYT2K3TMUXCXJ3C5H637366.json","view_paper":"https://pith.science/paper/4EXUYT2K","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.16832&json=true","fetch_graph":"https://pith.science/api/pith-number/4EXUYT2K3TMUXCXJ3C5H637366/graph.json","fetch_events":"https://pith.science/api/pith-number/4EXUYT2K3TMUXCXJ3C5H637366/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4EXUYT2K3TMUXCXJ3C5H637366/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4EXUYT2K3TMUXCXJ3C5H637366/action/storage_attestation","attest_author":"https://pith.science/pith/4EXUYT2K3TMUXCXJ3C5H637366/action/author_attestation","sign_citation":"https://pith.science/pith/4EXUYT2K3TMUXCXJ3C5H637366/action/citation_signature","submit_replication":"https://pith.science/pith/4EXUYT2K3TMUXCXJ3C5H637366/action/replication_record"}},"created_at":"2026-07-05T08:46:22.011237+00:00","updated_at":"2026-07-05T08:46:22.011237+00:00"}