{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:Z7RM57C26X55JISBIODA745SFY","short_pith_number":"pith:Z7RM57C2","schema_version":"1.0","canonical_sha256":"cfe2cefc5af5fbd4a24143860ff3b22e019cf6ca59ffa961e497032f972ae1d3","source":{"kind":"arxiv","id":"2607.22043","version":1},"attestation_state":"computed","paper":{"title":"Scaling Native Multimodal Pre-Training From Scratch","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Aoqi Wu, Bei Yu, Hai Wang, Haoyuan Wu, Jiajia Wu, Jinxiang Ou","submitted_at":"2026-07-24T07:13:52Z","abstract_excerpt":"Although large language models (LLMs) exhibit remarkable reasoning capabilities, their reliance on text-only pre-training restricts the perception of the multimodal physical world. Native multimodal pre-training avoids this limitation by training models from scratch on multimodal inputs, thereby achieving deep cross-modal integration and mitigating optimization asymmetries inherent to traditional late-fusion architectures. Despite these advantages, the scaling properties of this paradigm remain systematically uncharacterized. To address this gap, we investigate the optimal model size and token"},"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":"2607.22043","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-07-24T07:13:52Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"9a08260c0542d4941d1215f5a93942b7ff7a3b65cbb7c121ba2ed02939f095cb","abstract_canon_sha256":"6705d6d95bb56fd26be7583403abe9b849de6abad4823907822e553cb16cc347"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-27T01:20:44.884476Z","signature_b64":"iXpJMTUOiRBlkV5wNwACXrkKvlSO/zT4Qe3Ytmaemvu7BQjDf4Z065NtGfreEFQI7eAjLhlua8xZSE1g+OnvAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cfe2cefc5af5fbd4a24143860ff3b22e019cf6ca59ffa961e497032f972ae1d3","last_reissued_at":"2026-07-27T01:20:44.883568Z","signature_status":"signed_v1","first_computed_at":"2026-07-27T01:20:44.883568Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Native Multimodal Pre-Training From Scratch","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Aoqi Wu, Bei Yu, Hai Wang, Haoyuan Wu, Jiajia Wu, Jinxiang Ou","submitted_at":"2026-07-24T07:13:52Z","abstract_excerpt":"Although large language models (LLMs) exhibit remarkable reasoning capabilities, their reliance on text-only pre-training restricts the perception of the multimodal physical world. Native multimodal pre-training avoids this limitation by training models from scratch on multimodal inputs, thereby achieving deep cross-modal integration and mitigating optimization asymmetries inherent to traditional late-fusion architectures. Despite these advantages, the scaling properties of this paradigm remain systematically uncharacterized. To address this gap, we investigate the optimal model size and token"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.22043","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/2607.22043/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":"2607.22043","created_at":"2026-07-27T01:20:44.884023+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.22043v1","created_at":"2026-07-27T01:20:44.884023+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.22043","created_at":"2026-07-27T01:20:44.884023+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z7RM57C26X55","created_at":"2026-07-27T01:20:44.884023+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z7RM57C26X55JISB","created_at":"2026-07-27T01:20:44.884023+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z7RM57C2","created_at":"2026-07-27T01:20:44.884023+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.05000","citing_title":"Towards Physics of Multimodal Pretraining: Knowledge Flow, Modality Synergy, Early Unification, and Recipes","ref_index":137,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Z7RM57C26X55JISBIODA745SFY","json":"https://pith.science/pith/Z7RM57C26X55JISBIODA745SFY.json","graph_json":"https://pith.science/api/pith-number/Z7RM57C26X55JISBIODA745SFY/graph.json","events_json":"https://pith.science/api/pith-number/Z7RM57C26X55JISBIODA745SFY/events.json","paper":"https://pith.science/paper/Z7RM57C2"},"agent_actions":{"view_html":"https://pith.science/pith/Z7RM57C26X55JISBIODA745SFY","download_json":"https://pith.science/pith/Z7RM57C26X55JISBIODA745SFY.json","view_paper":"https://pith.science/paper/Z7RM57C2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.22043&json=true","fetch_graph":"https://pith.science/api/pith-number/Z7RM57C26X55JISBIODA745SFY/graph.json","fetch_events":"https://pith.science/api/pith-number/Z7RM57C26X55JISBIODA745SFY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z7RM57C26X55JISBIODA745SFY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z7RM57C26X55JISBIODA745SFY/action/storage_attestation","attest_author":"https://pith.science/pith/Z7RM57C26X55JISBIODA745SFY/action/author_attestation","sign_citation":"https://pith.science/pith/Z7RM57C26X55JISBIODA745SFY/action/citation_signature","submit_replication":"https://pith.science/pith/Z7RM57C26X55JISBIODA745SFY/action/replication_record"}},"created_at":"2026-07-27T01:20:44.884023+00:00","updated_at":"2026-07-27T01:20:44.884023+00:00"}