{"url":"/dataset/wit","name":"WIT","full_name":"Wikipedia-based Image Text","description_markdown":"**Wikipedia-based Image Text** (**WIT**) Dataset is a large multimodal multilingual dataset. WIT is composed of a curated set of 37.6 million entity rich image-text examples with 11.5 million unique images across 108 Wikipedia languages. Its size enables WIT to be used as a pretraining dataset for multimodal machine learning models.\r\n\r\n**Key Advantages**\r\n\r\nA few unique advantages of WIT:\r\n\r\n- The largest multimodal dataset (time of this writing) by the number of image-text examples.\r\n- A massively multilingual (first of its kind) with coverage for over 100+ languages.\r\n- A collection of diverse set of concepts and real world entities.\r\n- Brings forth challenging real-world test sets.","description_withheld":null,"homepage":"https://github.com/google-research-datasets/wit","introduced_date":"2021-03-02","introduced_date_note":null,"introduced_by":{"paper":"/paper/wit-wikipedia-based-image-text-dataset-for","title":"WIT: Wikipedia-based Image Text Dataset for Multimodal Multilingual Machine Learning","first_author":"Krishna Srinivasan","url":null},"license":null,"modalities":[{"name":"Images","url":"/datasets/modality/images"},{"name":"Texts","url":"/datasets/modality/texts"}],"tasks":[{"name":"Image Retrieval","url":"/task/image-retrieval","datasets_with_task":"/datasets/task/image-retrieval"}],"languages":[{"name":"Multilingual","url":"/datasets/language/multilingual"}],"variants":["WIT"],"data_loaders":[{"repo":"https://github.com/huggingface/datasets","url":"https://huggingface.co/datasets/wikimedia/wit_base","frameworks":["tf","pytorch","jax"]},{"repo":"https://github.com/huggingface/datasets","url":"https://huggingface.co/datasets/google/wit","frameworks":["tf","pytorch","jax"]},{"repo":"https://github.com/tensorflow/datasets","url":"https://www.tensorflow.org/datasets/catalog/wit","frameworks":["tf","pytorch","jax"]},{"repo":"https://github.com/google-research-datasets/wit","url":"https://github.com/google-research-datasets/wit","frameworks":[]}],"num_papers_in_archive":79,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[{"leaderboard":"/sota/image-retrieval-on-wit","task":"Image Retrieval","dataset_variant":"WIT","rows":2,"metrics":["R@1","R@5"],"first_row_in_archive_order":{"model":"WIT-ALL","paper":"/paper/wit-wikipedia-based-image-text-dataset-for","metrics":{"R@1":"0.346","R@5":"0.642"},"code_links":[{"title":"google-research-datasets/wit","url":"https://github.com/google-research-datasets/wit"},{"title":"clip-italian/clip-italian","url":"https://github.com/clip-italian/clip-italian"},{"title":"paullerner/viquae","url":"https://github.com/paullerner/viquae"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"}],"papers_with_a_benchmark_row":[{"paper":"/paper/wit-wikipedia-based-image-text-dataset-for","title":"WIT: Wikipedia-based Image Text Dataset for Multimodal Multilingual Machine Learning","date":"2021-03-02","rows_on_this_dataset":2,"code_links":3,"syntology":null}],"syntology_totals":{"read_at":"2026-09-24T18:15:14+00:00","papers_with_samples":0,"samples_harvested":0,"samples_ran":0,"samples_unverified":0,"pointer_only_for_licence":0,"papers_with_no_sample_that_ran":0,"note":"the per-paper counts above, summed; not a rate"},"papers_note":"The archive never published its papers-using-dataset list; these are papers with a leaderboard row on this dataset's benchmarks."}