Loading...
Loading...
News corpora, social media analysis, movie/music metadata, sports data, cultural datasets, misinformation
11,242 datasets
LiveBench releases new questions monthly to limit test set contamination for large language models. The benchmark includes questions derived from recently-released datasets, arXiv papers, news articles, and IMDb movie synopses. The dataset was created by 'livebench' and last updated in April 2025.
LiveBench Coding is a benchmark for Large Language Models containing approximately 1,000 coding-related questions designed to prevent test set contamination. Created by the LiveBench team and updated as recently as April 2025, the dataset utilizes objective ground-truth answers derived from recent sources like arXiv papers and news articles. It is released on a monthly cycle to ensure that the evaluation material remains outside the training window of most models.
A subset of Supervised Fine-Tuning (SFT) data used for training the Ring-lite-distill-preview language model. The dataset was created by inclusionAI and was last updated on the HuggingFace platform on April 15, 2025. Its specific size, format, and content details are not provided in the available metadata.
U.S. National Science Foundation social media data includes posts and content from platforms like Twitter, Facebook, LinkedIn, and YouTube. The dataset features tags for topics such as Engineering, Research, Funding, Science, and STEM careers. It was last updated in May 2025.
10 batches of Reddit text data annotated for self-disclosure using the CoNLL IOB2 sequence labeling format. Batches 1 through 8 contain single-annotator labels, while batches 9 and 10 provide higher-quality data through dual-annotation and expert adjudication.
A subset of Direct Preference Optimization data used to train the Ring-lite-distill-preview language model. The dataset was created by inclusionAI and was last updated on HuggingFace on April 15, 2025. Its specific size, format, and license details are not provided in the available metadata.
NFIRS data from Connecticut details fire department responses from 2012 to 2022, including fire, EMS, hazardous material, and false alarm incidents. The dataset includes columns for incident location, property use, casualties, resource deployment, and loss estimates. Data is voluntarily reported by fire departments to the national NFIRS program and released yearly by data.ct.gov with a considerable delay.
Chinese news articles summarized by an AI model. The dataset was created by author 'feilongfl' using ChatGPT 3.5 to process crawled news content, with summaries last updated on April 1, 2025. Each entry includes a title, summary, category, and timestamp.
A search interface for market news covering over 20 specific produce items including Broccoli, Apple, Celery, and Avocado. It is compiled by the Department of Agriculture's Fruit and Vegetable Program to aid producers, consumers, and distributors in national and international trade. The dataset was last updated in April 2025.
USDA's Fruit and Vegetable Market News Division provides information on fresh produce supplies, demand, and prices. The dataset includes searchable data for numerous commodities such as Broccoli, Apple, and Berries. It was last updated in April 2025.
BanTH is a dataset of 37.3 thousand transliterated Bangla text samples for hate speech detection and target identification. The data was sourced from 26 YouTube channels across News & Politics, People & Blogs, and Entertainment categories. It was created by aplycaebous and last updated on March 25, 2025.
Msp Podcast is an audio dataset published on the Hugging Face platform by CLAPv2. The dataset was last updated on 2025-05-11. Its specific content, scale, and collection methodology are not detailed in the available metadata.
A report on the activities of state, public libraries, and centralized library systems (CBS) in the Ivano-Frankivsk region of Ukraine. The dataset contains data on the number of libraries, readers, library workers, and library funds. It was published by the Department of Culture, Nationalities and Religions of the Ivano-Frankivsk Regional State Administration and last updated on 2025-04 04.
Lyrics for 12,751,073 music tracks were sourced from a database dump of LRCLIB on March 19, 2025. The dataset was uploaded by user ernestchu to Hugging Face and last updated on May 1, 2025. It is intended for text-based analysis and natural language processing tasks.
LiveBench is a benchmark designed to limit test set contamination by releasing new questions monthly. Questions are based on recently-released datasets, arXiv papers, news articles, and IMDb movie synopses. Each question has verifiable, objective ground-truth answers, allowing hard questions to be scored.
Regularly updated data from the OpenReview platform, annotated with metadata. The dataset is maintained by author 'nhop' and was last updated on April 2, 2025. Full paper texts are not included due to licensing but can be retrieved using provided code.
Bittensor Subnet 13 provides a continuously updated stream of preprocessed Reddit data from a decentralized network of miners. The dataset is maintained by the author 'tensorshield' and was last updated on 2025-04-02. Its real-time nature supports various analytical and machine learning tasks.
Six cultural variants of the GSM8K test set were created to investigate how cultural context influences large language model performance on mathematical problems. The dataset is part of a research project detailed in the paper 'Lost in Cultural Translation: Do LLMs Struggle with Math Across Cultural Contexts?' and was authored by abedk. It was last updated on Hugging Face in March 2025.
KidzRizal published a dataset of Twitter posts related to the Indonesian President on Hugging Face on 2025-05-13. The dataset likely contains text data for sentiment analysis. Metadata is minimal; actual content requires verification after download.
DurLAR is a high-fidelity 128-channel LiDAR dataset. It includes an exemplar subset of 600 frames available for preview, featuring image-based topics such as image_00, image_01, reflectivity, and ambient data. The dataset is intended for autonomous driving and robotics applications.