{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:EQOTML5SWCR3JLEFP6Y2I7MUYD","short_pith_number":"pith:EQOTML5S","schema_version":"1.0","canonical_sha256":"241d362fb2b0a3b4ac857fb1a47d94c0e2e0e9a0b80465c3960d090889f0c388","source":{"kind":"arxiv","id":"2504.07951","version":4},"attestation_state":"computed","paper":{"title":"Scaling Laws for Native Multimodal Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alaaeldin El-Nouby, Enrico Fini, Joshua Susskind, Matthieu Cord, Mustafa Shukor, Victor Guilherme Turrisi da Costa","submitted_at":"2025-04-10T17:57:28Z","abstract_excerpt":"Building general-purpose models that can effectively perceive the world through multimodal signals has been a long-standing goal. Current approaches involve integrating separately pre-trained components, such as connecting vision encoders to LLMs and continuing multimodal training. While such approaches exhibit remarkable sample efficiency, it remains an open question whether such late-fusion architectures are inherently superior. In this work, we revisit the architectural design of native multimodal models (NMMs)-those trained from the ground up on all modalities-and conduct an extensive scal"},"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":"2504.07951","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-04-10T17:57:28Z","cross_cats_sorted":[],"title_canon_sha256":"aac4f553b05e0b271be3a431467290c205b447e2eab508a71ff3b73f5c5a2912","abstract_canon_sha256":"15406706f337bd03f4256954ec12c224d60a8f584cc2b5b1faf2d411220bf2ec"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:51:25.283829Z","signature_b64":"OXRMSDFBYdLak3hJLGHkEAl/TIvxZfpCcRiS8nxPNJC/olCXr0LXO6j/AuyYS9ACwJm9m7fGALuHhq28pNtbAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"241d362fb2b0a3b4ac857fb1a47d94c0e2e0e9a0b80465c3960d090889f0c388","last_reissued_at":"2026-07-05T11:51:25.283335Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:51:25.283335Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Laws for Native Multimodal Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alaaeldin El-Nouby, Enrico Fini, Joshua Susskind, Matthieu Cord, Mustafa Shukor, Victor Guilherme Turrisi da Costa","submitted_at":"2025-04-10T17:57:28Z","abstract_excerpt":"Building general-purpose models that can effectively perceive the world through multimodal signals has been a long-standing goal. Current approaches involve integrating separately pre-trained components, such as connecting vision encoders to LLMs and continuing multimodal training. While such approaches exhibit remarkable sample efficiency, it remains an open question whether such late-fusion architectures are inherently superior. In this work, we revisit the architectural design of native multimodal models (NMMs)-those trained from the ground up on all modalities-and conduct an extensive scal"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.07951","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/2504.07951/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":"2504.07951","created_at":"2026-07-05T11:51:25.283398+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.07951v4","created_at":"2026-07-05T11:51:25.283398+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.07951","created_at":"2026-07-05T11:51:25.283398+00:00"},{"alias_kind":"pith_short_12","alias_value":"EQOTML5SWCR3","created_at":"2026-07-05T11:51:25.283398+00:00"},{"alias_kind":"pith_short_16","alias_value":"EQOTML5SWCR3JLEF","created_at":"2026-07-05T11:51:25.283398+00:00"},{"alias_kind":"pith_short_8","alias_value":"EQOTML5S","created_at":"2026-07-05T11:51:25.283398+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.01907","citing_title":"Lifting Unlabeled Internet-level Data for 3D Scene Understanding","ref_index":98,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10500","citing_title":"Visual Enhanced Depth Scaling for Multimodal Latent Reasoning","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08560","citing_title":"ZAYA1-VL-8B Technical Report","ref_index":93,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10500","citing_title":"Visual Enhanced Depth Scaling for Multimodal Latent Reasoning","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2506.01844","citing_title":"SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18827","citing_title":"OmniMouse: Scaling properties of multi-modal, multi-task Brain Models on 150B Neural Tokens","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10500","citing_title":"Visual Enhanced Depth Scaling for Multimodal Latent Reasoning","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07546","citing_title":"On the Invariance and Generality of Neural Scaling Laws","ref_index":46,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EQOTML5SWCR3JLEFP6Y2I7MUYD","json":"https://pith.science/pith/EQOTML5SWCR3JLEFP6Y2I7MUYD.json","graph_json":"https://pith.science/api/pith-number/EQOTML5SWCR3JLEFP6Y2I7MUYD/graph.json","events_json":"https://pith.science/api/pith-number/EQOTML5SWCR3JLEFP6Y2I7MUYD/events.json","paper":"https://pith.science/paper/EQOTML5S"},"agent_actions":{"view_html":"https://pith.science/pith/EQOTML5SWCR3JLEFP6Y2I7MUYD","download_json":"https://pith.science/pith/EQOTML5SWCR3JLEFP6Y2I7MUYD.json","view_paper":"https://pith.science/paper/EQOTML5S","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.07951&json=true","fetch_graph":"https://pith.science/api/pith-number/EQOTML5SWCR3JLEFP6Y2I7MUYD/graph.json","fetch_events":"https://pith.science/api/pith-number/EQOTML5SWCR3JLEFP6Y2I7MUYD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EQOTML5SWCR3JLEFP6Y2I7MUYD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EQOTML5SWCR3JLEFP6Y2I7MUYD/action/storage_attestation","attest_author":"https://pith.science/pith/EQOTML5SWCR3JLEFP6Y2I7MUYD/action/author_attestation","sign_citation":"https://pith.science/pith/EQOTML5SWCR3JLEFP6Y2I7MUYD/action/citation_signature","submit_replication":"https://pith.science/pith/EQOTML5SWCR3JLEFP6Y2I7MUYD/action/replication_record"}},"created_at":"2026-07-05T11:51:25.283398+00:00","updated_at":"2026-07-05T11:51:25.283398+00:00"}