Loading...
Loading...
Text classification, translation, QA, summarization, dialogue, sentiment analysis, language modeling, text corpora
46,236 datasets
SenseNova-U1 Training Sample is a small illustrative dataset created by sensenova. It exists solely to smoke-test the SenseNova-U1 training pipeline end-to-end. The dataset contains a handful of samples for every task type defined in the training configuration.
Synthetically generated conversational pairs created by Syntropy-AI for Project Syntropic. The dataset is intended to teach an AI model named Soren about its identity and conversational style. It serves as an early-stage personality and identity foundation layer for supervised fine-tuning.
A synthetic translation dataset distilled from multiple open corpora including WikiMatrix, TED2020, UNPC, ParaCrawl, and tldr-pages. Created by Neurora using a locally-hosted teacher model, it was last updated on May 25, 2026. The dataset is designed for edge deployment on Android devices to enable offline, low-latency machine translation.
LongAV-Compass provides prompt and event-level evaluation data for three multimodal generation tasks. The dataset covers text-to-audio-video, image-to-audio-video, and video-to-audio-video generation. It was authored by TengfeiLiuCoder and last updated on Hugging Face in May 2026.
RV Investigator voyage IN2025_V02 collected imagery of benthic habitats in the Huon Marine Park between March 26 and April 14, 2025. The dataset includes video and stereo stills captured by a Deep Towed camera, alongside navigational data. It was contributed by the Australian Ocean Data Network.
NOAA Ocean Exploration conducted a 21-day expedition in October-November 2021, focusing on the Blake Plateau and Florida Straits. The mission completed 14 remotely operated vehicle dives, totaling 47 hours of bottom time, and mapped 15,800 square kilometers of seafloor. Data from this expedition is archived and publicly available through NOAA.
14 remotely operated vehicle dives were conducted during a 21-day NOAA expedition in October-November 2021. Operations covered 15,800 square kilometers of seafloor mapping and collected 47 hours of bottom time video from depths between 300 and 3650 meters. The National Oceanic and Atmospheric Administration led this exploration of the Blake Plateau and Florida Straits.
AMOS (AMDSP Online Mapping and Quality Control System) is a web-based automated data validation tool used by Alberta municipalities. It validates GIS data such as roads, addresses, and common places against agreed municipal standards. The dataset is maintained by the Government of Alberta and was last updated on April 17, 2026.
Checklists from the Dutch government's content tools system, providing guidance for personal situations. The data is provided by the Ministry of the Interior and Kingdom Relations (Ministerie van Binnenlandse Zaan en Koninkrijksrelaties) under a CC0-1.0 license and is accessible via a JSON API. The last update date for the dataset is unknown.
Hawaii's Kaneohe Bay hosts this dataset from the National Renewable Energy Laboratory's Modular Ocean Instrumentation System (MOIS). It contains data files from the grid-connected deployment of the Northwest Energy Innovations (NWEI) Azura wave energy converter at the 30-meter berth of the U.S. Navy's Wave Energy Test Site. The dataset documents the October 2015 deployment to support marine hydrokinetic energy research.
A partial cache of 3,955,000 processed samples containing DC-AE image latents and Gemma text hidden states derived from JourneyDB-style image-caption pairs. Created by akrao9, this dataset is intended to reduce frozen encoder costs during Boomer model training. The cache generation was halted prematurely, skipping 849 source samples.
The dataset contains information about the network of sports grounds in the city of Kiev. It includes object names, addresses, geographical coordinates, administrative district assignments, balance holder data, object area, and accessibility information for persons with disabilities. The data originates from the ICS 'System of automation of relationships between participants in the sports educational process in the city of Kyiv' and is published by the States site of Ukraine.
Annual statistics for mineral exploration in Newfoundland and Labrador, likely containing data on mineral claims in good standing, mineral claims staked, diamond drilling meterage, and exploration expenditures. The dataset is published by the Government of Newfoundland and Labrador and was last updated on April 17, 2026.
Spatial-DISE is a benchmark dataset designed to evaluate spatial reasoning capabilities in vision-language models. It focuses on aspects of spatial intelligence including 3D perception, spatial transformation, and geometric reasoning across multiple difficulty levels. The dataset was created by TACPS-liv and was last updated on May 9, 2026.
Consolidated from January 2018 to December 2025, this dataset contains water quality measurements from a distribution network. It is published on the Socrata platform by the organization www.datos.gov.co. The data includes columns for parameters such as NITRITOS, E-COLI, PH, and TURBIEDAD.
Source data for Figure 1 of "Born-Qualified: An Autonomous Framework for Deploying Advanced Energy and Electronic Materials." The 17.8 KB XLSX file, authored by Steven Spurgeon and last updated on 2026-05-01, highlights market penetration and technological maturity for photovoltaics, batteries, and power electronics.
Columns suggest a dataset tracking complaints about pedestrian ramps and corners, likely containing records of service requests, repair feasibility assessments, and temporary repair actions. The dataset appears on multiple platforms, indicating its importance for urban infrastructure monitoring. Its structure includes identifiers for complaints and corners, dates for complaints and repairs, and details on repair types and feasibility.
Wenchao Ma's dataset systematically analyzes material metabolism, environmental leakage, and greenhouse gas emissions for 14 plastic types in China from 1992 to 2021. It models synergistic reduction potentials and cost-effectiveness under 14 future scenarios from 2021 to 2060. The data is hosted on figshare and was last updated in April 2026.
Force production and muscle activation measurements were collected from previously untrained individuals undergoing a six-week resistance training program. The dataset includes variables like peak force, rate of force development, rate of EMG rise, and intermuscular coherence, evaluated every two weeks. Researcher Paulo D. G. Santos published this data on figshare.
Yukon Energy Corporation provides data on electrical power substation locations within Yukon, Canada. The dataset details sites where voltage is transformed, typically from high to low. It is maintained by the Government of Yukon and receives updates when new substations are constructed.