{"url":"/sota/reading-comprehension-on-adversarialqa","task":{"name":"Reading Comprehension","url":"/task/reading-comprehension","note":null},"dataset":{"name":"AdversarialQA","url":"/dataset/adversarialqa"},"category":"Natural Language Processing","categories":["Natural Language Processing"],"category_note":null,"description":"Most current question answering datasets frame the task as reading comprehension where the question is about a paragraph or document and the answer often is a span in the document. \r\n\r\nSome specific tasks of reading comprehension include multi-modal machine reading comprehension and textual machine reading comprehension, among others. In the literature, machine reading comprehension can be divide into four categories: **cloze style**, **multiple choice**, **span prediction**, and **free-form answer**.  Read more about each category [here](https://paperswithcode.com/paper/a-survey-on-machine-reading-comprehension-1).\r\n\r\nBenchmark datasets used for testing a model's reading comprehension abilities include [MovieQA](/dataset/movieqa), [ReCoRD](dataset/record), and [RACE](/dataset/race), among others.\r\n\r\nThe Machine Reading group at UCL also provides an [overview of reading comprehension tasks](https://uclnlp.github.io/ai4exams/data.html).\r\n\r\nFigure source: [A Survey on Machine Reading Comprehension: Tasks, Evaluation Metrics and Benchmark Datasets](https://arxiv.org/pdf/2006.11880.pdf)","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Overall: F1","D(BiDAF): F1","D(BERT): F1","D(RoBERTa): F1"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Overall: F1":"higher","D(BiDAF): F1":"higher","D(BERT): F1":"higher","D(RoBERTa): F1":"higher"}},"counts":{"rows":3,"rows_with_code":3,"rows_with_paper_page":3,"rows_dated":3,"rows_using_additional_data":3},"rows":[{"rank_in_archive_order":1,"model":"RoBERTa-Large","metrics":{"D(BERT): F1":"65.5","D(BiDAF): F1":"74.1","D(RoBERTa): F1":"53.4","Overall: F1":"64.4"},"uses_additional_data":true,"paper_date":"2020-02-02","paper":"/paper/beat-the-ai-investigating-adversarial-human","paper_url":"https://arxiv.org/abs/2002.00293v2","paper_title":"Beat the AI: Investigating Adversarial Human Annotation for Reading Comprehension","code":"https://github.com/maxbartolo/adversarialQA","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"BERT-Large","metrics":{"D(BERT): F1":"62.4","D(BiDAF): F1":"71.3","D(RoBERTa): F1":"54.4","Overall: F1":"62.7"},"uses_additional_data":true,"paper_date":"2020-02-02","paper":"/paper/beat-the-ai-investigating-adversarial-human","paper_url":"https://arxiv.org/abs/2002.00293v2","paper_title":"Beat the AI: Investigating Adversarial Human Annotation for Reading Comprehension","code":"https://github.com/maxbartolo/adversarialQA","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"BiDAF","metrics":{"D(BERT): F1":"30.2","D(BiDAF): F1":"28.6","D(RoBERTa): F1":"26.7","Overall: F1":"28.5"},"uses_additional_data":true,"paper_date":"2020-02-02","paper":"/paper/beat-the-ai-investigating-adversarial-human","paper_url":"https://arxiv.org/abs/2002.00293v2","paper_title":"Beat the AI: Investigating Adversarial Human Annotation for Reading Comprehension","code":"https://github.com/maxbartolo/adversarialQA","n_code_links":1,"syntology":null}],"since_archive":{"present":false,"note":"No Syntology-extracted rows are published in this build."},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":0,"rows_with_any_sample_ran":0,"distinct_papers_with_graph_line":0,"distinct_papers_with_any_sample_ran":0,"samples_over_distinct_papers":{"n_ran":0,"n_unverified":0,"n_samples":0,"n_pointer_only_licence":0,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":0,"n_unverified":0,"n_samples":0,"n_pointer_only_licence":0,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}