{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:TTNY2ODJEGWATHHS3VDQJ3QXF2","short_pith_number":"pith:TTNY2ODJ","schema_version":"1.0","canonical_sha256":"9cdb8d386921ac099cf2dd4704ee172ea655464ac80816bd7db093177d342562","source":{"kind":"arxiv","id":"2205.01917","version":2},"attestation_state":"computed","paper":{"title":"CoCa: Contrastive Captioners are Image-Text Foundation Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"CoCa jointly trains contrastive and captioning losses in one encoder-decoder to create image-text foundation models that reach new state-of-the-art on ImageNet and multimodal tasks.","cross_cats":["cs.LG","cs.MM"],"primary_cat":"cs.CV","authors_text":"Jiahui Yu, Legg Yeung, Mojtaba Seyedhosseini, Vijay Vasudevan, Yonghui Wu, Zirui Wang","submitted_at":"2022-05-04T07:01:14Z","abstract_excerpt":"Exploring large-scale pretrained foundation models is of significant interest in computer vision because these models can be quickly transferred to many downstream tasks. This paper presents Contrastive Captioner (CoCa), a minimalist design to pretrain an image-text encoder-decoder foundation model jointly with contrastive loss and captioning loss, thereby subsuming model capabilities from contrastive approaches like CLIP and generative methods like SimVLM. In contrast to standard encoder-decoder transformers where all decoder layers attend to encoder outputs, CoCa omits cross-attention in 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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2205.01917","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2022-05-04T07:01:14Z","cross_cats_sorted":["cs.LG","cs.MM"],"title_canon_sha256":"979368834a9c1c63162fa1ef78acefae237399dabc836024cbfc08aa7f9b530b","abstract_canon_sha256":"33c64576bd99d3220fac09e2275b95093c5fe2928a27639a33ad2c3ad2b5f166"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:52.733484Z","signature_b64":"C5Pcho8EEjN+C+/arzBA+CER32b6WpIrC3Ht08S35YbnRDKr9RCu25CXCm0bFp9ES6oGP39X6/UhJ1h4R+PhBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9cdb8d386921ac099cf2dd4704ee172ea655464ac80816bd7db093177d342562","last_reissued_at":"2026-05-17T23:38:52.732937Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:52.732937Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CoCa: Contrastive Captioners are Image-Text Foundation Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"CoCa jointly trains contrastive and captioning losses in one encoder-decoder to create image-text foundation models that reach new state-of-the-art on ImageNet and multimodal tasks.","cross_cats":["cs.LG","cs.MM"],"primary_cat":"cs.CV","authors_text":"Jiahui Yu, Legg Yeung, Mojtaba Seyedhosseini, Vijay Vasudevan, Yonghui Wu, Zirui Wang","submitted_at":"2022-05-04T07:01:14Z","abstract_excerpt":"Exploring large-scale pretrained foundation models is of significant interest in computer vision because these models can be quickly transferred to many downstream tasks. This paper presents Contrastive Captioner (CoCa), a minimalist design to pretrain an image-text encoder-decoder foundation model jointly with contrastive loss and captioning loss, thereby subsuming model capabilities from contrastive approaches like CLIP and generative methods like SimVLM. In contrast to standard encoder-decoder transformers where all decoder layers attend to encoder outputs, CoCa omits cross-attention in the"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"CoCa obtains 86.3% zero-shot top-1 accuracy on ImageNet, 90.6% with a frozen encoder and learned classification head, and new state-of-the-art 91.0% top-1 accuracy on ImageNet with a finetuned encoder, while also leading on Kinetics, MSCOCO, VQA, and other tasks.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That omitting cross-attention in the first half of the decoder layers cleanly separates unimodal text representations from multimodal ones without harming overall capacity or optimization stability.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"CoCa unifies contrastive and generative pretraining in one image-text model to reach 86.3% zero-shot ImageNet accuracy and new state-of-the-art results on multiple downstream benchmarks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"CoCa jointly trains contrastive and captioning losses in one encoder-decoder to create image-text foundation models that reach new state-of-the-art on ImageNet and multimodal tasks.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"62f230de0b1498c62f4242da007033c4a6474eba7ea984251c75c03889628f78"},"source":{"id":"2205.01917","kind":"arxiv","version":2},"verdict":{"id":"1f439bdc-a6c5-418f-9c37-c73563009d9e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T10:48:13.746831Z","strongest_claim":"CoCa obtains 86.3% zero-shot top-1 accuracy on ImageNet, 90.6% with a frozen encoder and learned classification head, and new state-of-the-art 91.0% top-1 accuracy on ImageNet with a finetuned encoder, while also leading on Kinetics, MSCOCO, VQA, and other tasks.","one_line_summary":"CoCa unifies contrastive and generative pretraining in one image-text model to reach 86.3% zero-shot ImageNet accuracy and new state-of-the-art results on multiple downstream benchmarks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That omitting cross-attention in the first half of the decoder layers cleanly separates unimodal text representations from multimodal ones without harming overall capacity or optimization stability.","pith_extraction_headline":"CoCa jointly trains contrastive and captioning losses in one encoder-decoder to create image-text foundation models that reach new state-of-the-art on ImageNet and multimodal tasks."},"references":{"count":80,"sample":[{"doi":"","year":2021,"title":"On the Opportunities and Risks of Foundation Models","work_id":"a18039e9-928d-47c9-a836-32656a71bf71","ref_index":1,"cited_arxiv_id":"2108.07258","is_internal_anchor":true},{"doi":"","year":2018,"title":"BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding","work_id":"ed240a10-5b19-406c-baa5-30803f465785","ref_index":2,"cited_arxiv_id":"1810.04805","is_internal_anchor":true},{"doi":"","year":1910,"title":"Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer","work_id":"50e3b368-0243-4726-8186-233869802ad1","ref_index":3,"cited_arxiv_id":"1910.10683","is_internal_anchor":true},{"doi":"","year":1901,"title":"Language models are few-shot learners","work_id":"04bc68bc-b7df-4ec1-8599-da037bd4f085","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Thekkath, and Yonghui Wu","work_id":"6192fe03-df78-42e3-88fa-002af7197669","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":80,"snapshot_sha256":"15f6ce8074fb38f0cdc4f7262d52056ff86535f1620fd10356acbefe97446259","internal_anchors":15},"formal_canon":{"evidence_count":2,"snapshot_sha256":"ef2818b7fe9617f272021a3d4e2a136215e8d7bade34bf7398fd12bf60a5724f"},"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":"2205.01917","created_at":"2026-05-17T23:38:52.733023+00:00"},{"alias_kind":"arxiv_version","alias_value":"2205.01917v2","created_at":"2026-05-17T23:38:52.733023+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.01917","created_at":"2026-05-17T23:38:52.733023+00:00"},{"alias_kind":"pith_short_12","alias_value":"TTNY2ODJEGWA","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_16","alias_value":"TTNY2ODJEGWATHHS","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_8","alias_value":"TTNY2ODJ","created_at":"2026-05-18T12:33:33.725879+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":41,"internal_anchor_count":41,"sample":[{"citing_arxiv_id":"2405.07459","citing_title":"DAPL: Integration of Positive and Negative Descriptions in Text-Based Person Search","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2502.18816","citing_title":"Grad-ECLIP: Gradient-based Visual and Textual Explanations for CLIP","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2505.16819","citing_title":"Character-Centered Dialogue Generation from Scene-Level Prompts","ref_index":66,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21832","citing_title":"FLUID: From Ephemeral IDs to Multimodal Semantic Codes for Industrial-Scale Livestreaming Recommendation","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"2605.22098","citing_title":"TextTeacher: What Can Language Teach About Images?","ref_index":72,"is_internal_anchor":true},{"citing_arxiv_id":"2508.04227","citing_title":"Continual Learning for VLMs: A Survey and Taxonomy Beyond Forgetting","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2510.22665","citing_title":"SARVLM: A Vision Language Foundation Model for Semantic Understanding in SAR Imagery","ref_index":45,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20390","citing_title":"STELLAR: Scaling 3D Perception Large Models for Autonomous Driving","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21059","citing_title":"Multimodal LLMs under Pairwise Modalities","ref_index":69,"is_internal_anchor":true},{"citing_arxiv_id":"2605.16638","citing_title":"TTE-Flash: Accelerating Reasoning-based Multimodal Representations via Think-Then-Embed Tokens","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2605.16834","citing_title":"Learning Relative Representations for Fine-Grained Multimodal Alignment with Limited Data","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2511.21740","citing_title":"A cross-species neural foundation model for end-to-end speech decoding","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2212.03191","citing_title":"InternVideo: General Video Foundation Models via Generative and Discriminative Learning","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2512.17326","citing_title":"Democratising Pathology Co-Pilots: An Open Pipeline and Dataset for Whole-Slide Vision-Language Modelling","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2205.14100","citing_title":"GIT: A Generative Image-to-text Transformer for Vision and Language","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2303.15343","citing_title":"Sigmoid Loss for Language Image Pre-Training","ref_index":56,"is_internal_anchor":true},{"citing_arxiv_id":"2209.06794","citing_title":"PaLI: A Jointly-Scaled Multilingual Language-Image Model","ref_index":88,"is_internal_anchor":true},{"citing_arxiv_id":"2309.16671","citing_title":"Demystifying CLIP Data","ref_index":68,"is_internal_anchor":true},{"citing_arxiv_id":"2602.00937","citing_title":"CLAMP: Contrastive Learning for 3D Multi-View Action-Conditioned Robotic Manipulation Pretraining","ref_index":56,"is_internal_anchor":true},{"citing_arxiv_id":"2404.18416","citing_title":"Capabilities of Gemini Models in Medicine","ref_index":176,"is_internal_anchor":true},{"citing_arxiv_id":"2311.03079","citing_title":"CogVLM: Visual Expert for Pretrained Language Models","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2311.04400","citing_title":"LRM: Large Reconstruction Model for Single Image to 3D","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2302.12192","citing_title":"Aligning Text-to-Image Models using Human Feedback","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2312.14238","citing_title":"InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks","ref_index":170,"is_internal_anchor":true},{"citing_arxiv_id":"2210.08402","citing_title":"LAION-5B: An open large-scale dataset for training next generation image-text models","ref_index":92,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TTNY2ODJEGWATHHS3VDQJ3QXF2","json":"https://pith.science/pith/TTNY2ODJEGWATHHS3VDQJ3QXF2.json","graph_json":"https://pith.science/api/pith-number/TTNY2ODJEGWATHHS3VDQJ3QXF2/graph.json","events_json":"https://pith.science/api/pith-number/TTNY2ODJEGWATHHS3VDQJ3QXF2/events.json","paper":"https://pith.science/paper/TTNY2ODJ"},"agent_actions":{"view_html":"https://pith.science/pith/TTNY2ODJEGWATHHS3VDQJ3QXF2","download_json":"https://pith.science/pith/TTNY2ODJEGWATHHS3VDQJ3QXF2.json","view_paper":"https://pith.science/paper/TTNY2ODJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2205.01917&json=true","fetch_graph":"https://pith.science/api/pith-number/TTNY2ODJEGWATHHS3VDQJ3QXF2/graph.json","fetch_events":"https://pith.science/api/pith-number/TTNY2ODJEGWATHHS3VDQJ3QXF2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TTNY2ODJEGWATHHS3VDQJ3QXF2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TTNY2ODJEGWATHHS3VDQJ3QXF2/action/storage_attestation","attest_author":"https://pith.science/pith/TTNY2ODJEGWATHHS3VDQJ3QXF2/action/author_attestation","sign_citation":"https://pith.science/pith/TTNY2ODJEGWATHHS3VDQJ3QXF2/action/citation_signature","submit_replication":"https://pith.science/pith/TTNY2ODJEGWATHHS3VDQJ3QXF2/action/replication_record"}},"created_at":"2026-05-17T23:38:52.733023+00:00","updated_at":"2026-05-17T23:38:52.733023+00:00"}