{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:HIOPOGBL5KBH2MMS3452RWFKCO","short_pith_number":"pith:HIOPOGBL","schema_version":"1.0","canonical_sha256":"3a1cf7182bea827d3192df3ba8d8aa13a563c9bc399f26d39512261f5250fcd9","source":{"kind":"arxiv","id":"2502.07617","version":2},"attestation_state":"computed","paper":{"title":"Scaling Pre-training to One Hundred Billion Data for Vision Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Daniel Salz, Ibrahim Alabdulmohsin, Keran Rong, Xiaohua Zhai, Xiao Wang, Zhe Li","submitted_at":"2025-02-11T15:05:33Z","abstract_excerpt":"We provide an empirical investigation of the potential of pre-training vision-language models on an unprecedented scale: 100 billion examples. We find that model performance tends to saturate at this scale on many common Western-centric classification and retrieval benchmarks, such as COCO Captions. Nevertheless, tasks of cultural diversity achieve more substantial gains from the 100-billion scale web data, thanks to its coverage of long-tail concepts. Furthermore, we analyze the model's multilinguality and show gains in low-resource languages as well. In addition, we observe that reducing the"},"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":"2502.07617","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-02-11T15:05:33Z","cross_cats_sorted":[],"title_canon_sha256":"0024a178d8630d3977835be800f4c6e5806810f5bc6dd357723d7a62cf3c9bdb","abstract_canon_sha256":"a3f9bb6445095ea6824ee5d067ceae8b2492a8a35a3012dcad8b5f96493d2d33"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:04:05.124152Z","signature_b64":"fnCTbhWt6EJ2T8uP2fd9eQbHxoSWvTKNukZF4PTwQgLwuWaXY1siR5mrhv84flO2Rku9qOtGolsgV7sbBuntBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3a1cf7182bea827d3192df3ba8d8aa13a563c9bc399f26d39512261f5250fcd9","last_reissued_at":"2026-06-02T02:04:05.123664Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:04:05.123664Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Pre-training to One Hundred Billion Data for Vision Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Daniel Salz, Ibrahim Alabdulmohsin, Keran Rong, Xiaohua Zhai, Xiao Wang, Zhe Li","submitted_at":"2025-02-11T15:05:33Z","abstract_excerpt":"We provide an empirical investigation of the potential of pre-training vision-language models on an unprecedented scale: 100 billion examples. We find that model performance tends to saturate at this scale on many common Western-centric classification and retrieval benchmarks, such as COCO Captions. Nevertheless, tasks of cultural diversity achieve more substantial gains from the 100-billion scale web data, thanks to its coverage of long-tail concepts. Furthermore, we analyze the model's multilinguality and show gains in low-resource languages as well. In addition, we observe that reducing the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.07617","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/2502.07617/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":"2502.07617","created_at":"2026-06-02T02:04:05.123717+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.07617v2","created_at":"2026-06-02T02:04:05.123717+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.07617","created_at":"2026-06-02T02:04:05.123717+00:00"},{"alias_kind":"pith_short_12","alias_value":"HIOPOGBL5KBH","created_at":"2026-06-02T02:04:05.123717+00:00"},{"alias_kind":"pith_short_16","alias_value":"HIOPOGBL5KBH2MMS","created_at":"2026-06-02T02:04:05.123717+00:00"},{"alias_kind":"pith_short_8","alias_value":"HIOPOGBL","created_at":"2026-06-02T02:04:05.123717+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":6,"sample":[{"citing_arxiv_id":"2509.23370","citing_title":"GRAPE: Let GRPO Supervise Query Rewriting by Ranking for Retrieval","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2506.20670","citing_title":"MMSearch-R1: Incentivizing LMMs to Search","ref_index":62,"is_internal_anchor":true},{"citing_arxiv_id":"2603.23607","citing_title":"LongTail Driving Scenarios with Reasoning Traces: The KITScenes LongTail Dataset","ref_index":70,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09271","citing_title":"Shaping Schema via Language Representation as the Next Frontier for LLM Intelligence Expanding","ref_index":102,"is_internal_anchor":true},{"citing_arxiv_id":"2605.03547","citing_title":"Erase Persona, Forget Lore: Benchmarking Multimodal Copyright Unlearning in Large Vision Language Models","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2604.13313","citing_title":"Concrete Jungle: Towards Concreteness Paved Contrastive Negative Mining for Compositional Understanding","ref_index":8,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HIOPOGBL5KBH2MMS3452RWFKCO","json":"https://pith.science/pith/HIOPOGBL5KBH2MMS3452RWFKCO.json","graph_json":"https://pith.science/api/pith-number/HIOPOGBL5KBH2MMS3452RWFKCO/graph.json","events_json":"https://pith.science/api/pith-number/HIOPOGBL5KBH2MMS3452RWFKCO/events.json","paper":"https://pith.science/paper/HIOPOGBL"},"agent_actions":{"view_html":"https://pith.science/pith/HIOPOGBL5KBH2MMS3452RWFKCO","download_json":"https://pith.science/pith/HIOPOGBL5KBH2MMS3452RWFKCO.json","view_paper":"https://pith.science/paper/HIOPOGBL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.07617&json=true","fetch_graph":"https://pith.science/api/pith-number/HIOPOGBL5KBH2MMS3452RWFKCO/graph.json","fetch_events":"https://pith.science/api/pith-number/HIOPOGBL5KBH2MMS3452RWFKCO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HIOPOGBL5KBH2MMS3452RWFKCO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HIOPOGBL5KBH2MMS3452RWFKCO/action/storage_attestation","attest_author":"https://pith.science/pith/HIOPOGBL5KBH2MMS3452RWFKCO/action/author_attestation","sign_citation":"https://pith.science/pith/HIOPOGBL5KBH2MMS3452RWFKCO/action/citation_signature","submit_replication":"https://pith.science/pith/HIOPOGBL5KBH2MMS3452RWFKCO/action/replication_record"}},"created_at":"2026-06-02T02:04:05.123717+00:00","updated_at":"2026-06-02T02:04:05.123717+00:00"}