Loading...
Loading...
Text classification, translation, QA, summarization, dialogue, sentiment analysis, language modeling, text corpora
47,013 datasets
3,834 supervised fine-tuning samples accompany the paper 'When Vision Speaks for Sound'. This data mix, created by Rakancorle1, is designed to teach models to verify audio streams rather than rely on visual cues. It was last updated on May 19, 2026.
A dataset of Cornish-type engine house sites across Cornwall and west Devon, created for research purposes circa 2001-2002. It was compiled using historic and modern Ordnance Survey mapping alongside documentary research. The dataset includes extant structures with condition assessments and sites known only from map or documentary references.
Shujvkuyuanli is a Chinese text dataset authored by Beifengdo and hosted on Hugging Face. The dataset was last updated on July 7, 2026. Its specific content and size are not detailed in the available metadata.
Quaternary-era data describes the Mac Robertson shelf of East Antarctica, which was deeply eroded by glaciers and currents, exposing underlying basement. The dataset is provided by Geoscience Australia Data and was last updated on 2026-05-14. It likely contains information on the resulting 'scalped shelf' geomorphology.
Over 65,000 hours of human-transcribed speech across 127 language-region variants form this multilingual ASR dataset. The data is drawn from sources including national parliaments, public broadcasters, audiobooks, and international institutions. Each row consists of a 24 kHz speech utterance paired with a human transcript, an aligned ASR transcript, a character error rate, a WADA-SNR estimate, and four DNSMOS-P.835 quality scores.
45 sets of training data and five validation sets for predicting adsorption energies of carbon species on copper catalysts. The dataset includes a Jupyter notebook containing the machine learning model. Authored by Yuhui BAI and last updated in April 2026.
Australia's Identified Mineral Resources 2010 presents estimates of Australia's mineral resources at end of December 2009 for major and minor commodities. The report compares these long-term resource estimates with short-term industry ore reserves and includes analysis of mineral exploration expenditures for 2008-09. Data is compiled from Geoscience Australia, the Australian Bureau of Agricultural and Resource Economics and Sciences, and the United States Geological Survey.
9,253 parallel records translate Japanese to the endangered Ryukyuan (Okinawan) language, specifically the Shuri dialect. The dataset was structured from the 'Okinawan Language Dictionary' index published by the National Institute for Japanese Language and Linguistics (NINJAL) and uploaded by author 'doraking' in May 2026. It represents an extremely rare open data resource for NLP and machine learning, as UNESCO classifies Ryukyuan as a critically endangered language with an estimated speaker count below a few hundred.
A summary table of prior research investigating the impact of linguistic features on online advertising performance. The dataset was created by Kenjiro Inoue and last updated on April 15, 2026. It is a small 5.5 KB Excel file summarizing authors, platforms, linguistic variables, and key findings from seminal studies.
Results from a differential binding analysis of histone modification H3K18La across postnatal developmental time points P4, P10, and P28. The dataset contains statistical outputs from a quasi-likelihood negative binomial generalized log-linear model applied to normalized count data. It was authored by Mohita Gaur and uploaded to Figshare in April 2026.
Fifty-one diplomatic trips by President Barack Obama are analyzed using daily survey data to measure changes in domestic public approval. The dataset likely contains approval ratings before and after each trip, enabling analysis of short-term political effects. It was contributed by Benjamin Goldsmith to the International Studies Quarterly Dataverse and last updated in May 2026.
During the Second World War, buildings were sold by the Germans, the owners of which are mostly Jewish. The administration of these transactions is recorded in the Verkaufsbücher held by the National Archives. The records contain addresses and names of original owners, brokers, war buyers, and selling prices. In July 2019, a team from the National Archives and the Land Registry linked these transactions to the Kadaster's basic registration addresses and buildings (BAG) data, mapping the unlawful trades.
Geoscience Australia Data compiled a report incorporating data from geological reconnaissances in the Ramu Valley, the north-east foothills of the Schrader Ranges, and the Adelbert Mountains, including a brief visit to the Kumbrup area in the Simbai Valley. The report was last updated on 2026-05 06. It is provided in PDF and HTML formats.
Twenty-two de-identified interview transcripts from professional chaplains in the USA collected in early 2024. The data explores perspectives on expanding spiritual care into anonymous, asynchronous, and text-based online communities. The project was conducted by Alemitu Mequanint Bezabih and involves three research phases, with only Phase 1 data currently available.
Individual-level survey data from the Alberta Survey program captures determinants of health related to home ownership and security. The Government of Alberta collected this information via telephone surveys to support the creation of material and social deprivation indices. The dataset was last updated in April 2026.
Alberta Non-Profit Listing is a government registry of non-profit organizations incorporated or registered in the province. It includes organization type, name, status, registration date, city, and postal code. The Government of Alberta maintains the list, last updated in April 2026.
A dataset created for the Pseudoseer paper evaluation. It pairs search queries with target articles and their associated pseudocode blocks, including LaTeX source and reference context. The data was authored by Levent Toksoz and hosted on Harvard Dataverse, with a metadata update in June 2026.
Replication code for a study analyzing careers in finance using data from a major professional networking platform. The underlying data is confidential and not included, but the source code for reproducing the analysis is shared. The code was authored by Andrew Ellul and last updated in June 2026.
A collection of 64,691 samples for training specialized AI models in smart contract security. The dataset is structured for three tasks: vulnerability classification, security auditing with explanations and patches, and exploit proof-of-concept generation. It was created by aissacas and last updated on 2026-05-20.
Supplementary data from a 2026 study in Communications Earth & Environment provides a global compilation of whole-rock basalt geochemistry. The dataset tracks secular changes in CIPW normative mineralogy from the Archaean to the present day, supporting analysis of tectonic setting and mantle source evolution. It includes experimental melt compositions, LDA analysis outputs, and fractional crystallisation model results.