{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:DNKSUGZVW6QRAMY7S42SVDNA6F","short_pith_number":"pith:DNKSUGZV","schema_version":"1.0","canonical_sha256":"1b552a1b35b7a110331f97352a8da0f158e708038f9ead7d1bd639846c7d2926","source":{"kind":"arxiv","id":"2506.23115","version":1},"attestation_state":"computed","paper":{"title":"MoCa: Modality-aware Continual Pre-training Makes Better Bidirectional Multimodal Embeddings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Furu Wei, Haonan Chen, Hong Liu, Liang Wang, Nan Yang, Yuping Luo, Zhicheng Dou","submitted_at":"2025-06-29T06:41:00Z","abstract_excerpt":"Multimodal embedding models, built upon causal Vision Language Models (VLMs), have shown promise in various tasks. However, current approaches face three key limitations: the use of causal attention in VLM backbones is suboptimal for embedding tasks; scalability issues due to reliance on high-quality labeled paired data for contrastive learning; and limited diversity in training objectives and data. To address these issues, we propose MoCa, a two-stage framework for transforming pre-trained VLMs into effective bidirectional multimodal embedding models. The first stage, Modality-aware Continual"},"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":"2506.23115","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-06-29T06:41:00Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"c2cea8b7449aab276c8d0da647626191ea335d3a35db3dd5a28b33b46e2ba216","abstract_canon_sha256":"3cd55199acd0f73efd874c949cd4b02931d77807581fbba95343634f58091784"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:29:15.902574Z","signature_b64":"VW5bASJuF+Zip99oQTRK0TKyuB9KkbmS9jwNeuSQwAPv6/F/3ocoTEnstjX9k1wLwR1MVrPZiLVWiJkcHyjxAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1b552a1b35b7a110331f97352a8da0f158e708038f9ead7d1bd639846c7d2926","last_reissued_at":"2026-07-05T11:29:15.902036Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:29:15.902036Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MoCa: Modality-aware Continual Pre-training Makes Better Bidirectional Multimodal Embeddings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Furu Wei, Haonan Chen, Hong Liu, Liang Wang, Nan Yang, Yuping Luo, Zhicheng Dou","submitted_at":"2025-06-29T06:41:00Z","abstract_excerpt":"Multimodal embedding models, built upon causal Vision Language Models (VLMs), have shown promise in various tasks. However, current approaches face three key limitations: the use of causal attention in VLM backbones is suboptimal for embedding tasks; scalability issues due to reliance on high-quality labeled paired data for contrastive learning; and limited diversity in training objectives and data. To address these issues, we propose MoCa, a two-stage framework for transforming pre-trained VLMs into effective bidirectional multimodal embedding models. The first stage, Modality-aware Continual"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.23115","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/2506.23115/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":"2506.23115","created_at":"2026-07-05T11:29:15.902099+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.23115v1","created_at":"2026-07-05T11:29:15.902099+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.23115","created_at":"2026-07-05T11:29:15.902099+00:00"},{"alias_kind":"pith_short_12","alias_value":"DNKSUGZVW6QR","created_at":"2026-07-05T11:29:15.902099+00:00"},{"alias_kind":"pith_short_16","alias_value":"DNKSUGZVW6QRAMY7","created_at":"2026-07-05T11:29:15.902099+00:00"},{"alias_kind":"pith_short_8","alias_value":"DNKSUGZV","created_at":"2026-07-05T11:29:15.902099+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07654","citing_title":"MM-Matryoshka: Towards Budget-Elastic Visual Document Retrieval via a 2D Multimodal Matryoshka Training Framework","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21832","citing_title":"FLUID: From Ephemeral IDs to Multimodal Semantic Codes for Industrial-Scale Livestreaming Recommendation","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27295","citing_title":"Gemini Embedding 2: A Native Multimodal Embedding Model from Gemini","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21832","citing_title":"FLUID: From Ephemeral IDs to Multimodal Semantic Codes for Industrial-Scale Livestreaming Recommendation","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2509.18095","citing_title":"MetaEmbed: Scaling Multimodal Retrieval at Test-Time with Flexible Late Interaction","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02073","citing_title":"PLUME: Latent Reasoning Based Universal Multimodal Embedding","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06460","citing_title":"MINER: Mining Multimodal Internal Representation for Efficient Retrieval","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22280","citing_title":"Beyond Chain-of-Thought: Rewrite as a Universal Interface for Generative Multimodal Embeddings","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DNKSUGZVW6QRAMY7S42SVDNA6F","json":"https://pith.science/pith/DNKSUGZVW6QRAMY7S42SVDNA6F.json","graph_json":"https://pith.science/api/pith-number/DNKSUGZVW6QRAMY7S42SVDNA6F/graph.json","events_json":"https://pith.science/api/pith-number/DNKSUGZVW6QRAMY7S42SVDNA6F/events.json","paper":"https://pith.science/paper/DNKSUGZV"},"agent_actions":{"view_html":"https://pith.science/pith/DNKSUGZVW6QRAMY7S42SVDNA6F","download_json":"https://pith.science/pith/DNKSUGZVW6QRAMY7S42SVDNA6F.json","view_paper":"https://pith.science/paper/DNKSUGZV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.23115&json=true","fetch_graph":"https://pith.science/api/pith-number/DNKSUGZVW6QRAMY7S42SVDNA6F/graph.json","fetch_events":"https://pith.science/api/pith-number/DNKSUGZVW6QRAMY7S42SVDNA6F/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DNKSUGZVW6QRAMY7S42SVDNA6F/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DNKSUGZVW6QRAMY7S42SVDNA6F/action/storage_attestation","attest_author":"https://pith.science/pith/DNKSUGZVW6QRAMY7S42SVDNA6F/action/author_attestation","sign_citation":"https://pith.science/pith/DNKSUGZVW6QRAMY7S42SVDNA6F/action/citation_signature","submit_replication":"https://pith.science/pith/DNKSUGZVW6QRAMY7S42SVDNA6F/action/replication_record"}},"created_at":"2026-07-05T11:29:15.902099+00:00","updated_at":"2026-07-05T11:29:15.902099+00:00"}