Loading...
Loading...
Mathematical datasets, statistical benchmarks, probability, optimization, operations research
2,487 datasets
MathOlympiadBench contains human-verified formalizations of Olympiad-level mathematical competition problems. The dataset was created by Goedel-LM and introduced in a paper published in 2025. It sources problems from the Compfiles and IMOSLLean4 repository.
Statistical information on public information requests and citizen appeals processed by the Main Directorate of the State Tax Service in the Chernihiv region of Ukraine. The dataset includes reports on request satisfaction and work plan implementation, aggregated from the State site of Ukraine. It was last updated on August 4, 2025.
MEDLINE/PubMed annual statistical reports detail the content and size of data elements in the baseline versions of the database for 2018-2023. The reports include total citations and occurrences per element, plus minimum, average, and maximum occurrences and lengths. The data is provided by datadiscovery.nlm.nih.gov and was last updated on 2025-06-18.
Annual data from health.data.ny.gov reports statewide Licensed Home Care Services Agency (LHCSA) activity. It includes agency-level totals for patient counts, cases, and discharges for each reporting period. The dataset was last updated in July 2025.
Annual statistical reports detail the number of cases for 18 distinct home care services across New York State counties. The dataset is provided by the New York State Department of Health via health.data.ny.gov. Data was last updated in July 2025.
AutoMathText is a dataset of approximately 200 GB of mathematical texts compiled from sources including various websites, arXiv, and GitHub repositories like OpenWebMath, RedPajama, and Algebraic Stack. The dataset was created by author 'math-ai' and its associated work was accepted to ACL 2025 Findings. The dataset was last updated on July 16, 2025.
A strictly filtered subset of the NuminaMath-1.5 dataset containing only validated mathematical proof problems. The dataset was created by author 'nlile' and last updated on July 11, 2025. It applies multiple validation filters to the original data to ensure problem and solution validity.
MEDLINE/PubMed annual statistical reports detail the content and size of data elements within the biomedical literature database. The reports include counts of citations and element occurrences, plus minimum, average, and maximum values for occurrences and lengths per record. The data is provided by datadiscovery.nlm.nih.gov and was last updated on June 18, 2025.
MiroMind-M1 is a fully open-source series of reasoning language models built on Qwen-2.5. The dataset contains 719,000 curated problems used for supervised fine-tuning, with an additional 62,000 challenging examples used for reinforcement learning. It was created by miromind-ai and last updated on July 22, 2025.
Training performance data for the MiroMind-M1-RL-7B model on the AIME24 and AIME25 benchmarks. The dataset is associated with a model trained via reinforcement learning with verifiable rewards on 62,000 challenging examples. It is authored by miromind-ai and was last updated in July 2025.
Series of statistical reports on hazardous materials incidents compiled from the Hazardous Materials Incident Report Form 5800.1. The data is produced by the U.S. Department of Transportation and provides information on incidents by type, year, and geographical location. The latest update was in July 2025.
199 challenging mathematical problems designed to be at the limit of current LLM abilities. The dataset, named DAFT Math (Difficult Automatically-scorable Free-response Tasks for Math), was created by metr-evals and last updated on July 17, 2025. It is presented as a research artifact for a niche use-case.
Instance datasets of operations research problems. The repository, created by Oscar-Oliveira, was last updated on September 9, 2025. It contains curated datasets for classic optimization problems.
NVIDIA released this dataset in July 2025 containing fewer than 1,000 human-written solutions to complex math problems. It features extended reasoning chains authored by Olympiad-level mathematics students to emulate the step-by-step logic of advanced reasoning models.
A collection of mathematical formula pairs classified for equivalence. The dataset is based on 71 famous mathematical identities. It was created by author 'ddrg' and last updated on July 8, 2025.
132 geographic zones define U.S. domestic freight regions, classified as Metropolitan Areas, Remainder of State areas, or Whole States. The dataset originates from the 2017 Commodity Flow Survey and was published by the Bureau of Transportation Statistics in April 2022.
CombiBench consists of 100 manually produced combinatorial mathematics problems encoded in the Lean 4 formal language. Developed by AI-MO and updated in July 2025, it serves as a specialized benchmark for assessing the reasoning capabilities of automated theorem proving systems.
DeepTheorem is a framework for enhancing large language model mathematical reasoning through informal, natural language-based theorem proving. The dataset was created by Jiahao004 and last updated on Hugging Face on July 3, 2025. It introduces a novel approach to automated theorem proving by leveraging the informal reasoning strengths of LLMs.
NVIDIA's AceReason-Math contains 49,000 math problems and answers curated in June 2025 for training reasoning models via reinforcement learning. The collection is sourced from NuminaMath and DeepScaler-Preview, filtered to ensure all tasks are verifiable and text-based.
SupplyChainOptimization is a collection of documents discussing manufacturing processes. The dataset contains texts on logistics management, cost reduction, and demand forecasting methods. It was created by infinite-dataset-hub and last updated in June 2025.