Loading...
Loading...
Text classification, translation, QA, summarization, dialogue, sentiment analysis, language modeling, text corpora
46,225 datasets
Multi-angle Imaging SpectroRadiometer (MISR) Level 2 data contains stereoscopically derived cloud parameters from nine camera angles and four spectral bands. The dataset is produced by the National Aeronautics and Space Administration (NASA) to monitor monthly, seasonal, and long-term trends in aerosols, clouds, and land surface cover. It includes the Stereoscopically Derived Cloud Mask (SDCM), cloud winds, and Reflecting Level Reference Altitude (RLRA).
Anton Dohrn Seamount in UK waters features a geospatial map of Annex I and non-Annex I marine habitats, translated to the EUNIS classification system by the Joint Nature Conservation Committee (JNCC). Data was collected in July 2009 aboard the MV Franklin using multibeam echosounder acoustics and drop camera/towed video at 10 sites. The map includes polygons attributed with habitat types, biotopes, substrate classifications, and confidence scores based on the data sources used.
KFUPM-JRCAI's dataset contains machine-generated Arabic social media posts created using a text polishing approach across multiple Large Language Models. It was created for the research paper "Arabic machine-generated text detection: Stylometric analysis and cross-model evaluation". The dataset was last updated on May 22, 2026.
132.9 MB of Density Functional Theory (DFT) input and output files for calculating formation energies. The dataset, created by Lotte Theodora Janette Kortstee, supports a 2026 publication on strain-induced stability in double perovskite nanocrystals. It contains files for structural relaxations of bulk and surface cubic and trigonal Cs2Na1-xLixInCl6 phases.
From 2021-01-04 to 2026-03-31, this dataset provides a benchmarking corpus for contextual financial reasoning across 4,416 U.S. small- and micro-cap equities. It unifies seven tasks over a single point-in-time panel, including forecasting and valuation. The dataset was created by macrolens and last updated on the Hugging Face platform in May 2026.
A version of the LOFAR (Low Frequency Array) environmental vision adopted by the Provincial Staten van Drenthe on 2 July 2014. The dataset includes maps related to the telescope's location and regional planning, such as a vision map and a search area for large-scale wind energy. It was published by the Ministerie van Binnenlandse Zaken en Koninkrijksrelaties and is available as ZIP and PNG files under a CC-PDM-1.0 license.
January 2000 to November 2014 monthly snow cover frequency statistics for the Northern Hemisphere at a 1x1 degree spatial resolution. The dataset was prepared by Dr. Peter Romanov at the University of Maryland's Cooperative Institute for Climate Studies for the NEESPI program. Monthly data were derived from daily snow cover charts produced by NOAA/NESDIS.
Monthly ice cover frequency statistics for the Northern Hemisphere at a 1x1 degree spatial resolution. The dataset covers the time period from January 2000 to November 2014. It was prepared by Dr. Peter Romanov at the University of Maryland's Cooperative Institute for Climate Studies for the NEESPI program, derived from daily NOAA/NESDIS ice charts.
Melbourne's central business district is documented in 24 historical fire plan maps produced by Mahlstedt and Gee in 1888. The City of Melbourne digitized and geo-referenced these building plans, which are available for bulk download. The collection details building layouts and fire safety features from the late 19th century.
Electron Drift Instrument (EDI) data from the Cluster mission measures electric fields and magnetic field gradients in Earth's magnetosphere. The instrument emits a weak electron beam and measures its drift at a fundamental time step of 2 milliseconds. The dataset is provided by the National Aeronautics and Space Administration and was last updated in March 2026.
CLAP NQ is a benchmark dataset for evaluating the full Retrieval-Augmented Generation pipeline. It provides long answers grounded in gold passages from the Natural Questions dataset, with answers described as concise and cohesive. The dataset was created by PrimeQA and last updated on May 13, 2026.
January 2023 contract records from the municipality of Neiva, Colombia, published on the Socrata platform via datos.gov.co. The dataset includes columns for contract value, object, modality, and involved parties, suggesting a detailed view of local government procurement. The data was last updated on 2026-05-18.
NaviTrace is a Visual Question Answering benchmark designed to test how well vision-language models understand real-world navigation. It pairs first-person outdoor images with natural language instructions, requiring models to predict a 2D sequence of waypoints an agent would follow. The dataset was created by Voxel51 and was last updated on May 26, 2026.
Australia's Identified Mineral Resources is an annual nationwide assessment providing a long-term view of mineral resources available for mining. The 2011 revised edition includes Economic Demonstrated Resources (EDR), JORC Code ore reserves, and evaluations of long-term trends and world rankings. This assessment was published by the Australian Ocean Data Network and revised in 2014.
A 2026 study by Geoscience Australia established a thermal maturity framework for Permian sandstone reservoirs in the Northern Denison Trough, Bowen Basin, Australia. The dataset includes vitrinite reflectance measurements ranging from 0.55% to 0.93% and Rock-Eval Tmax values from 421°C to 447°C from coals and shales. It was created to evaluate CO2 storage potential and understand diagenetic processes affecting reservoir quality.
Water physics and chemistry data were collected over two days in February 1983 in the Northwest Atlantic Ocean. The dataset was submitted by the National Ocean Service and processed by the National Oceanographic Data Center to its standard F004 format. It contains measurements from moored current meters and bottle casts.
Experimental data from a six-week study tracks the maturation of bioprinted cartilaginous constructs. Measurements include glycosaminoglycan and collagen content alongside mechanical properties like hysteresis, shear modulus, nonlinearity, and stress relaxation at six time points. The dataset, authored by Paula Büttner, was published on figshare in 2026.
LongBench is a bilingual, multitask benchmark for evaluating long-context understanding in large language models. It covers scenarios including single-document QA, multi-document QA, summarization, few-shot learning, synthetic tasks, and code completion. The dataset was authored by GinkgoQ and repackaged for Hugging Face.
CNIL Sanctions Dataset contains 374 sanctions issued by the French data protection authority (CNIL) between 2011 and 2025. The dataset is structured into 34 analytical variables for quantitative research on GDPR and French privacy enforcement. It was generated by an autonomous text-to-dataset agent named Gemma Miner.
A bilingual English and Japanese judgment-eliciting Q&A corpus encodes the documented judgment from four research lines in the shimo4228 research program. The corpus is released under CC0 to maximize LLM-mediated diffusion and serves as an operational form of a specific authorship strategy tactic. It was authored by Shimo4228 and last updated on May 22, 2026.