Loading...
Loading...
Text classification, translation, QA, summarization, dialogue, sentiment analysis, language modeling, text corpora
44,940 datasets
ShahzebKhoso's repository hosts raw evaluation metrics and execution telemetry logs from running the Mostly Basic Python Problems (MBPP) benchmark against the Qwen 3.5 9B model. The data captures mid-tier performance and throughput dynamics of this generalist instruction model on local consumer hardware. The dataset was last updated on May 24, 2026.
EPM's monthly charges for natural gas service to households in Colombia from January 2020 to May 2026. The dataset includes a fixed charge and variable consumption charges, differentiated by socioeconomic stratum, with all values in Colombian pesos per cubic meter. Tariffs are calculated according to rules set by the Energy and Gas Regulatory Commission (CREG), specifically Resolution 011 of 2003.
The English Channel was surveyed from late February to mid-March 2012 as part of a Fisheries Science Partnership program. Cefas scientists observed 100 hauls to assess the viability of wreck netting for pollack, recording catch composition by species and weight. The work was funded by Defra and conducted in collaboration with the UK fishing industry.
A dataset of short Spanish conversations derived from thinkPy/oasst2-short-es. It includes two columns of automatically assigned topic classifications generated using the MoritzLaurer/mDeBERTa-v3-base-mnli-xnli model. The dataset was created by thinkPy and was last updated on 2026-05-31.
Protected Species Valuation, Phase One 2008 is a dataset from the National Oceanic and Atmospheric Administration (NOAA) and the Department of Commerce. It contains results from a stated preference choice experiment estimating the economic value of recovering or downlisting multiple threatened and endangered marine species in the United States. The research aims to synthesize over 40 existing nonmarket value estimates, which are often difficult to compare due to variations in study design.
Stream3D-1M-Dataset is a large-scale online spatio-temporal 3D question-answering dataset for training vision-language models. It contains over 1 million QA pairs generated from RGB-D video streams to support real-time 3D spatial understanding, temporal memory, and interactive reasoning. The dataset was introduced by author JonnyYu828 and was last updated on Hugging Face in June 2026.
South-east South Australia's stranded coastal barrier sequence preserves a record of sea-level variations over the past 800,000 years. This dataset presents new single-aliquot regenerative-dose optically stimulated luminescence (SAR-OSL) ages for quartz extracts from these dunes, testing the method's accuracy over a 0-250 ka range against an existing chronology. The data, associated with the Australian Ocean Data Network, was last updated in April 2026.
Eight studies analyzing the cost-effectiveness of quadrivalent influenza vaccines (QIV) in the general European population since 2012 are synthesized. The review, authored by Bram Van Sassenbroeck, found QIVs were cost-effective in 18 out of 19 analyses at a threshold of €35,000 per QALY. It was last updated in April 2026.
1841 to 1925 monthly mean magnetic data from the Greenwich Magnetic Observatory, digitized by the British Geological Survey. The dataset includes standardized inclination and declination values in arcminutes, with some horizontal and vertical force values converted to nanoTeslas. It provides a digital record of historical yearbooks for the magnetic research community.
Ai4Privacy's dataset contains 426,660 entries of personal financial information annotated for Personally Identifiable Information (PII). It includes 2,587,698 PII annotations across 38 label types, 30 languages, and 37 regions. The dataset is part of the larger PII-Masking-3M corpus and was last updated on June 3, 2026.
418,580 text entries with 2,305,517 annotations for personally identifiable information (PII) across 30 languages and 37 regions, created by Ai4Privacy. The dataset is part of the larger PII-Masking-3M corpus and was last updated on June 3, 2026.
Public art data from the municipal collection of Montreal, Quebec, distributed online via the Public Art site. The dataset likely contains information on works including title, artist, completion date, category, acquisition details, materials, district, location, and geolocation coordinates. It is provided by the Government and Municipalities of Québec under a CC-BY-4.0 license and was last updated on April 17, 2026.
Evaluation logs for four deep-research and web-agent benchmarks, published by i-DeepSearch on May 30, 2026. The repository contains judge results and trajectory logs from model runs, including a specific dataset named BrowseComp-Plus with 830 entries. Logs document performance under different observation mask context management settings.
A text generation corpus developed by dr-tkxx, last updated on 2026-06-02. The dataset is designed to support an AI model specialized in conceptual cryptography, esoteric data processing, abstract quantum simulations, and complex puzzle resolution. It contains text data in multiple languages, including Portuguese and English.
A 1.7 MB PDF document from Fabon Ltd. provides a product description for biomass pellet burners. The document details the technology's design, benefits, and industrial applications. It was last updated in April 2026.
TASTE is a human-evaluation corpus containing panel rankings of generated images across aesthetic and faithfulness dimensions. The dataset includes per-image hallucination judgements and was released alongside the TASTE preference model. It was authored by purvanshi and last updated on Hugging Face in May 2026.
GrepSeek is a dataset of 10,000 trajectories for cold-start supervised fine-tuning of a Direct Corpus Interaction search agent. The agent answers questions by issuing Unix shell commands like grep and rg directly against a raw text corpus, coupling retrieval and reasoning. The dataset was created by alireza7 and uploaded to Hugging Face on 2026-05-29.
Tery Setiawan's dataset investigates relations between economic resources, basic psychological needs, and subjective well-being in Indonesia's flood-prone regions. The dataset is 1.1 MB and includes syntax files for variable calculation in SPSS and R. It was last updated on May 6, 2026.
San Francisco's Housing and Homelessness department provides data for its shelter reservation waitlist. The dataset likely contains waitlist IDs generated from client birthdates and initials, as described in the user instructions. It was last updated on March 22, 2026.
Government of Alberta data tracks the number of individuals receiving services through the Persons with Developmental Disabilities program. The dataset records caseloads by three service delivery models: Family Managed Services, community service providers, and government-operated facilities. Data collection began in June 2018 and the dataset was last updated in April 2026.