{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2019:O5RS42J2BZVIBTBJQKLZBOM65Y","short_pith_number":"pith:O5RS42J2","canonical_record":{"source":{"id":"1908.03557","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2019-08-09T17:57:13Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"5234c3e317aa210886baf31c58a8a99a1e5db2eff2ebcdaf17dc66ddf228c066","abstract_canon_sha256":"c4108cec385808259cbc3762a1f4d246c7d6a97b3f4ce268bfa4a7f0e7335100"},"schema_version":"1.0"},"canonical_sha256":"77632e693a0e6a80cc29829790b99eee011c2a9d3da4f0c21a99d74a7a34be4d","source":{"kind":"arxiv","id":"1908.03557","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1908.03557","created_at":"2026-05-17T23:38:49Z"},{"alias_kind":"arxiv_version","alias_value":"1908.03557v1","created_at":"2026-05-17T23:38:49Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.03557","created_at":"2026-05-17T23:38:49Z"},{"alias_kind":"pith_short_12","alias_value":"O5RS42J2BZVI","created_at":"2026-05-18T12:33:24Z"},{"alias_kind":"pith_short_16","alias_value":"O5RS42J2BZVIBTBJ","created_at":"2026-05-18T12:33:24Z"},{"alias_kind":"pith_short_8","alias_value":"O5RS42J2","created_at":"2026-05-18T12:33:24Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2019:O5RS42J2BZVIBTBJQKLZBOM65Y","target":"record","payload":{"canonical_record":{"source":{"id":"1908.03557","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2019-08-09T17:57:13Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"5234c3e317aa210886baf31c58a8a99a1e5db2eff2ebcdaf17dc66ddf228c066","abstract_canon_sha256":"c4108cec385808259cbc3762a1f4d246c7d6a97b3f4ce268bfa4a7f0e7335100"},"schema_version":"1.0"},"canonical_sha256":"77632e693a0e6a80cc29829790b99eee011c2a9d3da4f0c21a99d74a7a34be4d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:49.690628Z","signature_b64":"5YY3+HB03qoERCefcoapWqQFyOjPDD7FPoNLha68ggPTQVb/h19fYOvHg3St+stnzuADOIChT4HIx4GBHU/jDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"77632e693a0e6a80cc29829790b99eee011c2a9d3da4f0c21a99d74a7a34be4d","last_reissued_at":"2026-05-17T23:38:49.690017Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:49.690017Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1908.03557","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:49Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"WxTF/D3vCZJEsoNXFmG1kDXMQ0aOSgzkuEuMZFEyj/3GmzSLel2usyLams28uLV3xSjgQ65i0wWNeBUHM4/5Ag==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-05T20:09:49.160957Z"},"content_sha256":"87cdb5d50e923b98a299070cac5c81b233a271ff7f6f8151bd4f591d939a18de","schema_version":"1.0","event_id":"sha256:87cdb5d50e923b98a299070cac5c81b233a271ff7f6f8151bd4f591d939a18de"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2019:O5RS42J2BZVIBTBJQKLZBOM65Y","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"VisualBERT: A Simple and Performant Baseline for Vision and Language","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"VisualBERT uses Transformer self-attention to align text tokens with image regions from caption data alone.","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.CV","authors_text":"Cho-Jui Hsieh, Da Yin, Kai-Wei Chang, Liunian Harold Li, Mark Yatskar","submitted_at":"2019-08-09T17:57:13Z","abstract_excerpt":"We propose VisualBERT, a simple and flexible framework for modeling a broad range of vision-and-language tasks. VisualBERT consists of a stack of Transformer layers that implicitly align elements of an input text and regions in an associated input image with self-attention. We further propose two visually-grounded language model objectives for pre-training VisualBERT on image caption data. Experiments on four vision-and-language tasks including VQA, VCR, NLVR2, and Flickr30K show that VisualBERT outperforms or rivals with state-of-the-art models while being significantly simpler. Further analy"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments on four vision-and-language tasks including VQA, VCR, NLVR2, and Flickr30K show that VisualBERT outperforms or rivals with state-of-the-art models while being significantly simpler.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That self-attention inside the Transformer layers can learn useful alignments between text tokens and image regions from caption data alone without any explicit grounding supervision.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"VisualBERT is a Transformer model that implicitly aligns text and image regions through self-attention and achieves competitive or superior results on VQA, VCR, NLVR2, and Flickr30K after pre-training on captions.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"VisualBERT uses Transformer self-attention to align text tokens with image regions from caption data alone.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"0fb539f8d28e15f9c4aa2ac7a4f17f1ab0ec814bce835043f0ea4e82a9cd57dd"},"source":{"id":"1908.03557","kind":"arxiv","version":1},"verdict":{"id":"de89ea96-5c53-4c7d-89b4-b0f19183eb3b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T23:56:11.486724Z","strongest_claim":"Experiments on four vision-and-language tasks including VQA, VCR, NLVR2, and Flickr30K show that VisualBERT outperforms or rivals with state-of-the-art models while being significantly simpler.","one_line_summary":"VisualBERT is a Transformer model that implicitly aligns text and image regions through self-attention and achieves competitive or superior results on VQA, VCR, NLVR2, and Flickr30K after pre-training on captions.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That self-attention inside the Transformer layers can learn useful alignments between text tokens and image regions from caption data alone without any explicit grounding supervision.","pith_extraction_headline":"VisualBERT uses Transformer self-attention to align text tokens with image regions from caption data alone."},"references":{"count":113,"sample":[{"doi":"","year":2018,"title":"Bottom-up and top-down attention for image captioning and visual question answering","work_id":"2a42d0f2-4f84-4b3f-a8e5-67625a794f1d","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2015,"title":"VQA : Visual question answering","work_id":"4d21e295-7618-4f2d-8a91-e11543796e93","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"MUREL : Multimodal relational reasoning for visual question answering","work_id":"f29e71ce-4071-44f2-acff-50f85bd13447","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"What does BERT look at? an analysis of BERT 's attention","work_id":"06202934-6df8-4956-a50e-d37fe9955c26","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2008,"title":"Stanford typed dependencies manual","work_id":"97860c0b-0d29-4241-b9f5-6d954ae90258","ref_index":6,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":113,"snapshot_sha256":"8fed92b8706ae7fdf98da700a47c6ba01c2fda339fe703b4e9a0adbf4553332b","internal_anchors":9},"formal_canon":{"evidence_count":2,"snapshot_sha256":"83d860fbc0d43f5ba073d0d0ee76f1ca582fff7fcb3ac0183dee6e26b79eb059"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"de89ea96-5c53-4c7d-89b4-b0f19183eb3b"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:49Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"8YotZeOnjQ6corUApQlaKVTFS1KGMYNFZH9tlIBItyqHUkNWxrZf5A9LepmeR/k6xN/mICvUuiwGwQnJg1poCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-05T20:09:49.161959Z"},"content_sha256":"a70a7f3d773792918cbafcae8dc6329144f59d0011eb4ce194a43feac2a99adc","schema_version":"1.0","event_id":"sha256:a70a7f3d773792918cbafcae8dc6329144f59d0011eb4ce194a43feac2a99adc"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/O5RS42J2BZVIBTBJQKLZBOM65Y/bundle.json","state_url":"https://pith.science/pith/O5RS42J2BZVIBTBJQKLZBOM65Y/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/O5RS42J2BZVIBTBJQKLZBOM65Y/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-05T20:09:49Z","links":{"resolver":"https://pith.science/pith/O5RS42J2BZVIBTBJQKLZBOM65Y","bundle":"https://pith.science/pith/O5RS42J2BZVIBTBJQKLZBOM65Y/bundle.json","state":"https://pith.science/pith/O5RS42J2BZVIBTBJQKLZBOM65Y/state.json","well_known_bundle":"https://pith.science/.well-known/pith/O5RS42J2BZVIBTBJQKLZBOM65Y/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:O5RS42J2BZVIBTBJQKLZBOM65Y","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"c4108cec385808259cbc3762a1f4d246c7d6a97b3f4ce268bfa4a7f0e7335100","cross_cats_sorted":["cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2019-08-09T17:57:13Z","title_canon_sha256":"5234c3e317aa210886baf31c58a8a99a1e5db2eff2ebcdaf17dc66ddf228c066"},"schema_version":"1.0","source":{"id":"1908.03557","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1908.03557","created_at":"2026-05-17T23:38:49Z"},{"alias_kind":"arxiv_version","alias_value":"1908.03557v1","created_at":"2026-05-17T23:38:49Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.03557","created_at":"2026-05-17T23:38:49Z"},{"alias_kind":"pith_short_12","alias_value":"O5RS42J2BZVI","created_at":"2026-05-18T12:33:24Z"},{"alias_kind":"pith_short_16","alias_value":"O5RS42J2BZVIBTBJ","created_at":"2026-05-18T12:33:24Z"},{"alias_kind":"pith_short_8","alias_value":"O5RS42J2","created_at":"2026-05-18T12:33:24Z"}],"graph_snapshots":[{"event_id":"sha256:a70a7f3d773792918cbafcae8dc6329144f59d0011eb4ce194a43feac2a99adc","target":"graph","created_at":"2026-05-17T23:38:49Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Experiments on four vision-and-language tasks including VQA, VCR, NLVR2, and Flickr30K show that VisualBERT outperforms or rivals with state-of-the-art models while being significantly simpler."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That self-attention inside the Transformer layers can learn useful alignments between text tokens and image regions from caption data alone without any explicit grounding supervision."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"VisualBERT is a Transformer model that implicitly aligns text and image regions through self-attention and achieves competitive or superior results on VQA, VCR, NLVR2, and Flickr30K after pre-training on captions."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"VisualBERT uses Transformer self-attention to align text tokens with image regions from caption data alone."}],"snapshot_sha256":"0fb539f8d28e15f9c4aa2ac7a4f17f1ab0ec814bce835043f0ea4e82a9cd57dd"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"83d860fbc0d43f5ba073d0d0ee76f1ca582fff7fcb3ac0183dee6e26b79eb059"},"paper":{"abstract_excerpt":"We propose VisualBERT, a simple and flexible framework for modeling a broad range of vision-and-language tasks. VisualBERT consists of a stack of Transformer layers that implicitly align elements of an input text and regions in an associated input image with self-attention. We further propose two visually-grounded language model objectives for pre-training VisualBERT on image caption data. Experiments on four vision-and-language tasks including VQA, VCR, NLVR2, and Flickr30K show that VisualBERT outperforms or rivals with state-of-the-art models while being significantly simpler. Further analy","authors_text":"Cho-Jui Hsieh, Da Yin, Kai-Wei Chang, Liunian Harold Li, Mark Yatskar","cross_cats":["cs.CL","cs.LG"],"headline":"VisualBERT uses Transformer self-attention to align text tokens with image regions from caption data alone.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2019-08-09T17:57:13Z","title":"VisualBERT: A Simple and Performant Baseline for Vision and Language"},"references":{"count":113,"internal_anchors":9,"resolved_work":113,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Bottom-up and top-down attention for image captioning and visual question answering","work_id":"2a42d0f2-4f84-4b3f-a8e5-67625a794f1d","year":2018},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"VQA : Visual question answering","work_id":"4d21e295-7618-4f2d-8a91-e11543796e93","year":2015},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"MUREL : Multimodal relational reasoning for visual question answering","work_id":"f29e71ce-4071-44f2-acff-50f85bd13447","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"What does BERT look at? an analysis of BERT 's attention","work_id":"06202934-6df8-4956-a50e-d37fe9955c26","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":6,"title":"Stanford typed dependencies manual","work_id":"97860c0b-0d29-4241-b9f5-6d954ae90258","year":2008}],"snapshot_sha256":"8fed92b8706ae7fdf98da700a47c6ba01c2fda339fe703b4e9a0adbf4553332b"},"source":{"id":"1908.03557","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T23:56:11.486724Z","id":"de89ea96-5c53-4c7d-89b4-b0f19183eb3b","model_set":{"reader":"grok-4.3"},"one_line_summary":"VisualBERT is a Transformer model that implicitly aligns text and image regions through self-attention and achieves competitive or superior results on VQA, VCR, NLVR2, and Flickr30K after pre-training on captions.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"VisualBERT uses Transformer self-attention to align text tokens with image regions from caption data alone.","strongest_claim":"Experiments on four vision-and-language tasks including VQA, VCR, NLVR2, and Flickr30K show that VisualBERT outperforms or rivals with state-of-the-art models while being significantly simpler.","weakest_assumption":"That self-attention inside the Transformer layers can learn useful alignments between text tokens and image regions from caption data alone without any explicit grounding supervision."}},"verdict_id":"de89ea96-5c53-4c7d-89b4-b0f19183eb3b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:87cdb5d50e923b98a299070cac5c81b233a271ff7f6f8151bd4f591d939a18de","target":"record","created_at":"2026-05-17T23:38:49Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"c4108cec385808259cbc3762a1f4d246c7d6a97b3f4ce268bfa4a7f0e7335100","cross_cats_sorted":["cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2019-08-09T17:57:13Z","title_canon_sha256":"5234c3e317aa210886baf31c58a8a99a1e5db2eff2ebcdaf17dc66ddf228c066"},"schema_version":"1.0","source":{"id":"1908.03557","kind":"arxiv","version":1}},"canonical_sha256":"77632e693a0e6a80cc29829790b99eee011c2a9d3da4f0c21a99d74a7a34be4d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"77632e693a0e6a80cc29829790b99eee011c2a9d3da4f0c21a99d74a7a34be4d","first_computed_at":"2026-05-17T23:38:49.690017Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:49.690017Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"5YY3+HB03qoERCefcoapWqQFyOjPDD7FPoNLha68ggPTQVb/h19fYOvHg3St+stnzuADOIChT4HIx4GBHU/jDw==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:49.690628Z","signed_message":"canonical_sha256_bytes"},"source_id":"1908.03557","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:87cdb5d50e923b98a299070cac5c81b233a271ff7f6f8151bd4f591d939a18de","sha256:a70a7f3d773792918cbafcae8dc6329144f59d0011eb4ce194a43feac2a99adc"],"state_sha256":"0d5f71cb8c178686353756b999a5b92964e4ccd0efbc81a9e36dde2928f82f57"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tkHQPcqbS7WS1NrkvmvIioiWcPXT2Ozdo4KoNBwu580P8oSso7SzxrCGG2puYoJF1B3mMgZrbboWsIpeFnR2Aw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-05T20:09:49.166763Z","bundle_sha256":"f83b2b639e899794b7ef4a53edc585bb5f8a6490982932df9110d070ee494a0d"}}