{"success":true,"data":{"author":"DeepMind (Ling et al.)","bucketAccessType":"public","bucketName":"deepmind/aqua_rat","bucketPrefix":"","category":"Mathematics","cloudProvider":"huggingface","createdAt":"2026-11-22T01:09:15.614Z","description":"AQuA-RAT (Algebra Question Answering with Rationales) is a set of about 100,000 algebraic and arithmetic word problems, each with five multiple-choice options, the correct option and a rationale that explains the solution step by step in natural language mixed with equations. It was built by DeepMind to train models that generate an explanation and a program that solves the problem, and is now a common benchmark for math word problem solving, quantitative reasoning and chain-of-thought rationale generation. Splits: 97,467 training, 254 validation and 254 test problems. The repository has two configurations as Parquet files: raw/ (question, options, rationale, correct answer) and tokenized/ (the same fields pre-tokenized), about 52 MB in total. Load with the Hugging Face datasets library (load_dataset('deepmind/aqua_rat', 'raw')) or pandas. The data is published in the deepmind/AQuA GitHub repository under the Apache License 2.0; cite Ling et al., Program Induction by Rationale Generation: Learning to Solve and Explain Algebraic Word Problems, ACL 2017.","displayTitle":"AQuA-RAT: 100,000 Algebraic Word Problems with Multiple-Choice Answers and Step-by-Step Natural Language Rationales (DeepMind)","huggingFaceUrl":"https://huggingface.co/datasets/deepmind/aqua_rat","id":"35445d54-2905-5bb1-97e8-c9d88ac6cc44","license":"Apache-2.0","licenseEvidenceUrl":"https://github.com/deepmind/AQuA/blob/master/LICENSE","objectCountEstimate":8,"probeStatus":"ok","referenceUrl":"https://arxiv.org/abs/1705.04146","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"deepmind/aqua_rat\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"deepmind/aqua_rat\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=deepmind, license:apache-2.0, size_categories:100K<n<1M, format:parquet, downloads=17068"],"format":"mixed","generatedAt":"2026-11-22T01:10:10.447Z","layout":{"example":"raw/test-00000-of-00001.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"50415231150415848a0615f4c1034c15fc031500120000828503f0753a0100004120636172206973206265696e672064726976656e2c20696e2061207374726169676874206c696e6520616e64206174206120756e69666f726d2073706565642c20746f7761726473207468652062617365206f66206120766572746963616c20746f7765722e2054686520746f70206f660529007401160177306f627365727665642066726f6d051b019008616e640584907468652070726f636573732c2069742074616b6573203130206d696e7574657320666f7220012510616e676c65017a48656c65766174696f6e20746f206368616e6765095b0c3435c2b00114a03630c2b02e20416674657220686f77206d756368206d6f72652074696d652077696c6c2074686973200188107265616368058600620dda016d05bd003f213e40546865206f726967696e616c2070726963058418616e206974656d01df78646973636f756e746564203232252e204120637573746f6d65722062757973054f052c04617409761d31095004757321985861202432302d6f666620636f75706f6e2e2054686572650163186e6f207461782001f2046865054d002c2140094f04776109640c6f6e6c79051c05721984206f756768742e20496636180024706169642024312e39302d1b1868616e2068616c092746f200197a0c776861741976421b","encoding":"hex","key":"raw/test-00000-of-00001.parquet","truncatedAt":512}],"sampleObjects":["raw/test-00000-of-00001.parquet","raw/train-00000-of-00001.parquet","raw/validation-00000-of-00001.parquet","tokenized/test-00000-of-00001.parquet","tokenized/train-00000-of-00001.parquet"],"sidecars":[{"bytes":"---\nannotations_creators:\n- crowdsourced\nlanguage_creators:\n- crowdsourced\n- expert-generated\nlanguage:\n- en\nlicense:\n- apache-2.0\nmultilinguality:\n- monolingual\nsize_categories:\n- 10K<n<100K\nsource_datasets:\n- original\ntask_categories:\n- question-answering\ntask_ids:\n- multiple-choice-qa\npaperswithcode_id: aqua-rat\npretty_name: Algebra Question Answering with Rationales\ndataset_info:\n- config_name: raw\n  features:\n  - name: question\n    dtype: string\n  - name: options\n    sequence: string\n  - name: rationale\n    dtype: string\n  - name: correct\n    dtype: string\n  splits:\n  - name: train\n    num_bytes: 42333059\n    num_examples: 97467\n  - name: test\n    num_bytes: 116759\n    num_examples: 254\n  - name: validation\n    num_bytes: 118616\n    num_examples: 254\n  download_size: 25568676\n  dataset_size: 42568434\n- config_name: tokenized\n  features:\n  - name: question\n    dtype: string\n  - name: options\n    sequence: string\n  - name: rationale\n    dtype: string\n  - name: correct\n    dtype: string\n  splits:\n  - name: train\n    num_bytes: 46493643\n    num_examples: 97467\n  - name: test\n    num_bytes: 126263\n    num_examples: 254\n  - name: validation\n    num_bytes: 128853\n    num_examples: 254\n  download_size: 26429873\n  dataset_size: 46748759\nconfigs:\n- config_name: raw\n  data_files:\n  - split: train\n    path: raw/train-*\n  - split: test\n    path: raw/test-*\n  - split: validation\n    path: raw/validation-*\n  default: true\n- config_name: tokenized\n  data_files:\n  - split: train\n    path: tokenized/train-*\n  - split: test\n    path: tokenized/test-*\n  - split: validation\n    path: tokenized/validation-*\n---\n\n# Dataset Card for AQUA-RAT\n\n## Table of Contents\n- [Dataset Description](#dataset-description)\n  - [Dataset Summary](#dataset-summary)\n  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)\n  - [Languages](#languages)\n- [Dataset Structure](#dataset-structure)\n  - [Data Instances](#data-instances)\n  - [Data Fields](#data-fields)\n  - [Data Splits](#data-splits)\n- [Dataset Creation](#dataset-creation)\n  - [Curation Rationale](#curation-rationale)\n  - [Source Data](#source-data)\n  - [Annotations](#annotations)\n  - [Personal and Sensitive Information](#personal-and-sensitive-information)\n- [Considerations for Using the Data](#considerations-for-using-the-data)\n  - [Social Impact of Dataset](#social-impact-of-dataset)\n  - [Discussion of Biases](#discussion-of-biases)\n  - [Other Known Limitations](#other-known-limitations)\n- [Additional Information](#additional-information)\n  - [Dataset Curators](#dataset-curators)\n  - [Licensing Information](#licensing-information)\n  - [Citation Information](#citation-information)\n  - [Contributions](#contributions)\n\n## Dataset Description\n\n- **Homepage:** [https://github.com/deepmind/AQuA](https://github.com/deepmind/AQuA)\n- **Repository:** [https://github.com/deepmind/AQuA](https://github.com/deepmind/AQuA)\n- **Paper:** [https://arxiv.org/pdf/1705.04146.pdf](https://arxiv.org/pdf/1705.04146.pdf)\n\n### Dataset Summary\n\nA large-scale dataset consisting of approximately 100,000 algebraic word problems.\nThe solution to each question is explained step-by-step using natural language.\nThis data is used to train a program generation model that learns to generate the explanation,\nwhile generating the program that solves the question.\n\n### Supported Tasks and Leaderboards\n\n### Languages\n\nen\n\n## Dataset Structure\n\n### Data Instances\n```\n{\n\"question\": \"A grocery sells a bag of ice for $1.25, and makes 20% profit. If it sells 500 bags of ice, how much total profit does it make?\",\n\"options\": [\"A)125\", \"B)150\", \"C)225\", \"D)250\", \"E)275\"],\n\"rationale\": \"Profit per bag = 1.25 * 0.20 = 0.25\\nTotal profit = 500 * 0.25 = 125\\nAnswer is A.\",\n\"correct\": \"A\"\n}\n```\n\n### Data Fields\n\n- `question` : (str) A natural language definition of the problem to solve\n- `options` : (list(str)) 5 possible options (A, B, C, D and E), among which one is correct\n- `rationale` : (str) A natural language description of the solution to the problem\n- `correct` : (str) The correct option\n\n### Data Splits\n|                    ","encoding":"utf8","key":"README.md","name":"README.md","size":5891,"truncated":true}],"sourceObjectKey":"raw/test-00000-of-00001.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"deepmind/aqua_rat\")\nprint(ds)","starterSamplePath":"raw/test-00000-of-00001.parquet","totalObjects":8},"summary":"AQuA-RAT (Algebra Question Answering with Rationales) is a set of about 100,000 algebraic and arithmetic word problems, each with five multiple-choice options, the correct option…","title":"AQuA-RAT: 100,000 Algebraic Word Problems with Multiple-Choice Answers and Step-by-Step Natural Language Rationales (DeepMind)","updatedAt":"2026-11-22T04:05:06.702Z","visibility":"public","websiteUrl":"https://github.com/deepmind/AQuA"}}