{"url":"/dataset/clevr","name":"CLEVR","full_name":"Compositional Language and Elementary Visual Reasoning","description_markdown":"**CLEVR** (**Compositional Language and Elementary Visual Reasoning**) is a synthetic Visual Question Answering dataset. It contains images of 3D-rendered objects; each image comes with a number of highly compositional questions that fall into different categories. Those categories fall into 5 classes of tasks: Exist, Count, Compare Integer, Query Attribute and Compare Attribute. The CLEVR dataset consists of: a training set of 70k images and 700k questions, a validation set of 15k images and 150k questions, a test set of 15k images and 150k questions about objects, answers, scene graphs and functional programs for all train and validation images and questions. Each object present in the scene, aside of position, is characterized by a set of four attributes: 2 sizes: large, small, 3 shapes: square, cylinder, sphere, 2 material types: rubber, metal, 8 color types: gray, blue, brown, yellow, red, green, purple, cyan, resulting in 96 unique combinations.\r\n\r\nSource: [On transfer learning using a MAC model variant](https://arxiv.org/abs/1811.06529)\r\nImage Source: [Johnson et al](https://arxiv.org/pdf/1612.06890v1.pdf)","description_withheld":null,"homepage":"https://cs.stanford.edu/people/jcjohns/clevr/","introduced_date":"2016-12-20","introduced_date_note":null,"introduced_by":{"paper":"/paper/clevr-a-diagnostic-dataset-for-compositional","title":"CLEVR: A Diagnostic Dataset for Compositional Language and Elementary Visual Reasoning","first_author":"Justin Johnson","url":null},"license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"modalities":[{"name":"Images","url":"/datasets/modality/images"},{"name":"Texts","url":"/datasets/modality/texts"}],"tasks":[{"name":"Image Generation","url":"/task/image-generation","datasets_with_task":"/datasets/task/image-generation"},{"name":"Visual Question Answering (VQA)","url":"/task/visual-question-answering","datasets_with_task":"/datasets/task/visual-question-answering"},{"name":"Visual Question Answering","url":"/task/visual-question-answering-1","datasets_with_task":"/datasets/task/visual-question-answering-1"},{"name":"Visual Question Answering (VQA) Split A","url":"/task/visual-question-answering-vqa-split-a","datasets_with_task":"/datasets/task/visual-question-answering-vqa-split-a"},{"name":"Visual Question Answering (VQA) Split B","url":"/task/visual-question-answering-vqa-split-b","datasets_with_task":"/datasets/task/visual-question-answering-vqa-split-b"}],"languages":[],"variants":["CLEVR","CLEVR-CoGenT"],"data_loaders":[{"repo":"https://github.com/pytorch/vision","url":"https://github.com/pytorch/vision","frameworks":["pytorch"]},{"repo":"https://github.com/facebookresearch/ParlAI","url":"https://parl.ai/docs/tasks.html#clevr","frameworks":["pytorch"]},{"repo":"https://github.com/tensorflow/datasets","url":"https://www.tensorflow.org/datasets/catalog/clevr","frameworks":["tf","jax"]}],"num_papers_in_archive":657,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[{"leaderboard":"/sota/visual-question-answering-on-clevr","task":"Visual Question Answering (VQA)","dataset_variant":"CLEVR","rows":15,"metrics":["Accuracy"],"first_row_in_archive_order":{"model":"NS-VQA (1K programs)","paper":"/paper/neural-symbolic-vqa-disentangling-reasoning","metrics":{"Accuracy":"99.8"},"code_links":[{"title":"kexinyi/ns-vqa","url":"https://github.com/kexinyi/ns-vqa"},{"title":"nerdimite/neuro-symbolic-ai-soc","url":"https://github.com/nerdimite/neuro-symbolic-ai-soc"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/image-generation-on-clevr","task":"Image Generation","dataset_variant":"CLEVR","rows":6,"metrics":["FID-5k-training-steps"],"first_row_in_archive_order":{"model":"Projected GAN","paper":"/paper/projected-gans-converge-faster","metrics":{"FID-5k-training-steps":"0.89"},"code_links":[{"title":"autonomousvision/projected_gan","url":"https://github.com/autonomousvision/projected_gan"},{"title":"dome272/ProjectedGAN-pytorch","url":"https://github.com/dome272/ProjectedGAN-pytorch"},{"title":"tsubota-kouga/ProjectedGAN","url":"https://github.com/tsubota-kouga/ProjectedGAN"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/visual-question-answering-on-clevr-1","task":"Visual Question Answering","dataset_variant":"CLEVR","rows":1,"metrics":["Accuracy"],"first_row_in_archive_order":{"model":"NeSyCoCo Neuro-Symbolic","paper":"/paper/nesycoco-a-neuro-symbolic-concept-composer","metrics":{"Accuracy":"99.7"},"code_links":[{"title":"hlr/nesycoco","url":"https://github.com/hlr/nesycoco"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"}],"papers_with_a_benchmark_row":[{"paper":"/paper/nesycoco-a-neuro-symbolic-concept-composer","title":"NeSyCoCo: A Neuro-Symbolic Concept Composer for Compositional Generalization","date":"2024-12-20","rows_on_this_dataset":2,"code_links":1,"syntology":null},{"paper":"/paper/projected-gans-converge-faster","title":"Projected GANs Converge Faster","date":"2021-11-01","rows_on_this_dataset":1,"code_links":3,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":49,"samples_ran":38,"samples_unverified":11,"pointer_only_for_licence":6,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/mdetr-modulated-detection-for-end-to-end","title":"MDETR -- Modulated Detection for End-to-End Multi-Modal Understanding","date":"2021-04-26","rows_on_this_dataset":1,"code_links":5,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":11,"samples_ran":6,"samples_unverified":5,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/generative-adversarial-transformers","title":"Generative Adversarial Transformers","date":"2021-03-01","rows_on_this_dataset":5,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":6,"samples_ran":5,"samples_unverified":1,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/interpretable-visual-reasoning-via-induced","title":"Interpretable Visual Reasoning via Induced Symbolic Space","date":"2020-11-23","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/language-conditioned-graph-networks-for","title":"Language-Conditioned Graph Networks for Relational Reasoning","date":"2019-05-10","rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":10,"samples_ran":0,"samples_unverified":10,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/the-neuro-symbolic-concept-learner-1","title":"The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision","date":"2019-04-26","rows_on_this_dataset":1,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":7,"samples_ran":0,"samples_unverified":7,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/explainable-and-explicit-visual-reasoning","title":"Explainable and Explicit Visual Reasoning over Scene Graphs","date":"2018-12-05","rows_on_this_dataset":1,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":3,"samples_ran":2,"samples_unverified":1,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/neural-symbolic-vqa-disentangling-reasoning","title":"Neural-Symbolic VQA: Disentangling Reasoning from Vision and Language Understanding","date":"2018-10-04","rows_on_this_dataset":1,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":2,"samples_ran":2,"samples_unverified":0,"pointer_only_for_licence":2,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/question-guided-hybrid-convolution-for-visual","title":"Question-Guided Hybrid Convolution for Visual Question Answering","date":"2018-08-08","rows_on_this_dataset":1,"code_links":0,"syntology":null},{"paper":"/paper/learning-visual-question-answering-by","title":"Learning Visual Question Answering by Bootstrapping Hard Attention","date":"2018-08-01","rows_on_this_dataset":1,"code_links":1,"syntology":null},{"paper":"/paper/ddrprog-a-clevr-differentiable-dynamic","title":"DDRprog: A CLEVR Differentiable Dynamic Reasoning Programmer","date":"2018-03-30","rows_on_this_dataset":1,"code_links":0,"syntology":null},{"paper":"/paper/transparency-by-design-closing-the-gap","title":"Transparency by Design: Closing the Gap Between Performance and Interpretability in Visual Reasoning","date":"2018-03-14","rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":8,"samples_ran":0,"samples_unverified":8,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/compositional-attention-networks-for-machine","title":"Compositional Attention Networks for Machine Reasoning","date":"2018-03-08","rows_on_this_dataset":1,"code_links":10,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":7,"samples_ran":1,"samples_unverified":6,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/film-visual-reasoning-with-a-general","title":"FiLM: Visual Reasoning with a General Conditioning Layer","date":"2017-09-22","rows_on_this_dataset":1,"code_links":7,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":10,"samples_ran":9,"samples_unverified":1,"pointer_only_for_licence":7,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/a-simple-neural-network-module-for-relational","title":"A simple neural network module for relational reasoning","date":"2017-06-05","rows_on_this_dataset":1,"code_links":20,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":7,"samples_ran":3,"samples_unverified":4,"pointer_only_for_licence":1,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/inferring-and-executing-programs-for-visual","title":"Inferring and Executing Programs for Visual Reasoning","date":"2017-05-10","rows_on_this_dataset":1,"code_links":5,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":6,"samples_ran":4,"samples_unverified":2,"pointer_only_for_licence":6,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}}],"syntology_totals":{"read_at":"2026-09-24T18:15:14+00:00","papers_with_samples":12,"samples_harvested":126,"samples_ran":70,"samples_unverified":56,"pointer_only_for_licence":22,"papers_with_no_sample_that_ran":3,"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."}