{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:SGSALVUPYCWNDFNZFCECHZGITQ","short_pith_number":"pith:SGSALVUP","schema_version":"1.0","canonical_sha256":"91a405d68fc0acd195b9288823e4c89c363b450089b848edcd040a06d585c5e3","source":{"kind":"arxiv","id":"2203.05175","version":1},"attestation_state":"computed","paper":{"title":"MVP: Multimodality-guided Visual Pre-training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Houqiang Li, Lingxi Xie, Longhui Wei, Qi Tian, Wengang Zhou","submitted_at":"2022-03-10T06:11:20Z","abstract_excerpt":"Recently, masked image modeling (MIM) has become a promising direction for visual pre-training. In the context of vision transformers, MIM learns effective visual representation by aligning the token-level features with a pre-defined space (e.g., BEIT used a d-VAE trained on a large image corpus as the tokenizer). In this paper, we go one step further by introducing guidance from other modalities and validating that such additional knowledge leads to impressive gains for visual pre-training. The proposed approach is named Multimodality-guided Visual Pre-training (MVP), in which we replace 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":"2203.05175","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2022-03-10T06:11:20Z","cross_cats_sorted":[],"title_canon_sha256":"528bd10a713d1b81f15e5847095106603ec00aaa68350c81ce7569f37826ba6d","abstract_canon_sha256":"bc0dfc2d4843c7f1a57b1bfa54375611b447edc792299d36f4de9cef36f745e4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:03:30.746752Z","signature_b64":"3azmqtisVoH/X84eQLl6dtthy4eBtwrz00aIeYQ44R7vsYw3r/XQz3eF5jw8dwcualVUgh8DruUbfZl5HHJGBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"91a405d68fc0acd195b9288823e4c89c363b450089b848edcd040a06d585c5e3","last_reissued_at":"2026-07-05T04:03:30.746335Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:03:30.746335Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MVP: Multimodality-guided Visual Pre-training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Houqiang Li, Lingxi Xie, Longhui Wei, Qi Tian, Wengang Zhou","submitted_at":"2022-03-10T06:11:20Z","abstract_excerpt":"Recently, masked image modeling (MIM) has become a promising direction for visual pre-training. In the context of vision transformers, MIM learns effective visual representation by aligning the token-level features with a pre-defined space (e.g., BEIT used a d-VAE trained on a large image corpus as the tokenizer). In this paper, we go one step further by introducing guidance from other modalities and validating that such additional knowledge leads to impressive gains for visual pre-training. The proposed approach is named Multimodality-guided Visual Pre-training (MVP), in which we replace the "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.05175","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/2203.05175/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":"2203.05175","created_at":"2026-07-05T04:03:30.746395+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.05175v1","created_at":"2026-07-05T04:03:30.746395+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.05175","created_at":"2026-07-05T04:03:30.746395+00:00"},{"alias_kind":"pith_short_12","alias_value":"SGSALVUPYCWN","created_at":"2026-07-05T04:03:30.746395+00:00"},{"alias_kind":"pith_short_16","alias_value":"SGSALVUPYCWNDFNZ","created_at":"2026-07-05T04:03:30.746395+00:00"},{"alias_kind":"pith_short_8","alias_value":"SGSALVUP","created_at":"2026-07-05T04:03:30.746395+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.31201","citing_title":"ExPLoRe: Expert Patch-Level Loss Routing for Multi-Objective Masked Image Modeling","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SGSALVUPYCWNDFNZFCECHZGITQ","json":"https://pith.science/pith/SGSALVUPYCWNDFNZFCECHZGITQ.json","graph_json":"https://pith.science/api/pith-number/SGSALVUPYCWNDFNZFCECHZGITQ/graph.json","events_json":"https://pith.science/api/pith-number/SGSALVUPYCWNDFNZFCECHZGITQ/events.json","paper":"https://pith.science/paper/SGSALVUP"},"agent_actions":{"view_html":"https://pith.science/pith/SGSALVUPYCWNDFNZFCECHZGITQ","download_json":"https://pith.science/pith/SGSALVUPYCWNDFNZFCECHZGITQ.json","view_paper":"https://pith.science/paper/SGSALVUP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.05175&json=true","fetch_graph":"https://pith.science/api/pith-number/SGSALVUPYCWNDFNZFCECHZGITQ/graph.json","fetch_events":"https://pith.science/api/pith-number/SGSALVUPYCWNDFNZFCECHZGITQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SGSALVUPYCWNDFNZFCECHZGITQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SGSALVUPYCWNDFNZFCECHZGITQ/action/storage_attestation","attest_author":"https://pith.science/pith/SGSALVUPYCWNDFNZFCECHZGITQ/action/author_attestation","sign_citation":"https://pith.science/pith/SGSALVUPYCWNDFNZFCECHZGITQ/action/citation_signature","submit_replication":"https://pith.science/pith/SGSALVUPYCWNDFNZFCECHZGITQ/action/replication_record"}},"created_at":"2026-07-05T04:03:30.746395+00:00","updated_at":"2026-07-05T04:03:30.746395+00:00"}