Loading...
Loading...
Text classification, translation, QA, summarization, dialogue, sentiment analysis, language modeling, text corpora
46,121 datasets
YEG 2010 contains up-to-date information on mining and mineral exploration activity, studies by industry, and results of recent geological field studies. Information in this volume comes from prospectors, exploration and government geologists, mining companies and students who are willing to contribute to public geoscience. The dataset is published by the Government of Yukon and was last updated in April 2026.
A government report detailing exploration and mining activity in the Yukon for the year 1987. Summaries are based on reports submitted to the department for assessment credit by mineral exploration companies. The report discusses the geology of Yukon mineral deposits and districts under active investigation.
1.9 million instruction pairs for ancient-to-modern and modern-to-ancient translation accompany a structured corpus of classical Chinese texts. The corpus includes the complete Thirteen Classics, the Shuowen Jiezi dictionary, the Comprehensive Mirror for Aid in Government, and the first 15 of the Twenty-Four Histories. Created by dzxr and released under CC0, it is available on HuggingFace with a last update timestamp of 2026-04-29.
Historical commodity prices from 2000 to 2010 are compiled from multiple government ministries and the US Federal Reserve. The data is presented in tabular format and published by the Government of British Columbia under an open license. It was last updated on April 17, 2026.
Texas Department of Insurance data on waiver requests from major medical health plans seeking exemptions from state network adequacy standards. It tracks insurer applications, hearing dates, and commissioner decisions for specific counties and provider specialties. The dataset is published by data.texas.gov and was last updated in March 2026.
SpatialForge-10M is a large-scale vision-language dataset designed for 3D-aware spatial perception and reasoning from open-world 2D images. It contains over 10 million question-answer pairs generated from 2.8 million curated images. The dataset was created by researchers from Lingnan University and XPENG Robotics, with the latest update recorded on 2026-05-25.
1982-2015 data on winning candidates, votes received, popular vote percentage, number of names on the list, and voter turnout percentage for each electoral division in Alberta general elections. The dataset is provided by the Government of Alberta and was last updated on 2026-04-17.
The MD_COLUMN_NAME_MAPPING table maps Victorian Spatial Data Library object column names longer than 10 characters to shorter names. This mapping supports the generation and importing of shapefiles, which have a 10-character column name limit. The Department of Transport and Planning maintains this reference table, last updated on 2026-04-09.
6kplus created this collection of 21,348 text prompts derived from MotionX. The prompts were used to train and evaluate the PhyMotion human-motion video-generation method, as detailed in the associated research paper.
5,000 high-quality reasoning traces distilled to mirror the step-by-step thinking of Llama 4 Maverick class models. The dataset was synthetically generated by Meta AI and published by WithinUsAI on May 24, 2026. It is designed for training student language models to reproduce a specific chain-of-thought pattern.
Yasmine N. Biré's dataset records seasonal variations in container productivity for Aedes aegypti mosquito pupae. The data covers peri-urban and intra-urban villages in the Cocody-Bingerville area of southeastern Côte d’Ivoire. Observations were collected over a one-year period from August 2023 to July 2024.
MotIF-1K is a robotics motion dataset containing 1,022 demonstrations across 13 task categories. It is used to benchmark and fine-tune vision-language models for motion-based success detection. The dataset was created by Voxel51 and was last updated on the Hugging Face platform in May 2026.
Characterization results from a materials science study include TEM, XRD, and FTIR analyses. The data comprises figures and raw/processed files supporting the structural, morphological, and functional properties of synthesized materials. The dataset is 921.1 KB in size and was authored by Ebrahim Saied, last updated in May 2026.
Lava flows interpreted from drill hole intersections and magnetic geophysical responses. The dataset is part of a collection of geological features for the Gippsland and Otway Basins in Victoria, Australia, produced by the Department of Energy, Environment and Climate Action. It was last updated in April 2026.
A 2025-2026 departmental plan from the Canadian Human Rights Commission, published on the open_canada platform. The document outlines planned spending, human resources, and core responsibilities including Engagement & Advocacy, Complaints, and Proactive Compliance. It is licensed under the OGL-CA-2.0 and was last updated in April 2026.
106 million tokens of synthetic documents generated for evaluating AI models. Documents were created using GPT-4.1 for brainstorming and GPT-5 for final generation, based on the pipeline from Wang et al. (2025). The dataset is organized by seven evaluation traits, including Verifiable Structure, Placeholder, and Ethical Dilemma.
Hyperspectral imagery provides calibrated surface reflectance measurements from the AVIRIS-NG airborne instrument. The dataset contains 72 flight lines captured over research sites in Idaho and California during 2014 and 2015. Data is radiometrically and geometrically calibrated, offering reflectance across 426 spectral bands at 1-meter spatial resolution.
A graticule layer for map-making software, consisting of lines spaced at 1-kilometer intervals attributed with Easting and Northing values. The dataset is provided by the Department of Energy, Environment and Climate Action and was last updated on 2026-04-09. It is part of a series of layers for common map projections and can be reprojected to overlay data in other coordinate systems.
13,203 training samples provide token-level annotations for hallucinated spans in LLM-generated code responses to real developer tasks from SWE-bench. The dataset, created by KRLabsOrg and last updated in May 2026, uses the unified LettuceDetect v2 taxonomy for character-level labeling. It includes separate splits for training, development, and testing, with 5,039 samples explicitly labeled as hallucinated in the training set.
RoboStressBench is a robotics-oriented vision-language benchmark for evaluating model robustness under visual stress conditions. It contains two task formats: visual question answering and localization tasks. The dataset was created by RoboStressBench and was last updated on June 2, 2026.