Loading...
Loading...
News corpora, social media analysis, movie/music metadata, sports data, cultural datasets, misinformation
11,234 datasets
UTSD aggregates time series data across 7 distinct domains, containing up to 1 billion individual time points. The dataset is structured into four hierarchical volumes to support research on large models and pre-training. It was created by the thuml organization and updated in 2024 and 2025.
State of California provides grant award data for Fiscal Year 2022-2023, mandated by AB132 for grantmakers. The dataset contains post-award information submitted by grantmakers for awards closing on or after July 1, 2022.
High-quality summaries of BBC news articles from the XSum dataset, generated using the Llama-4-Maverick-17B-128E-Instruct-FP8 model. The dataset was created by PursuitOfDataScience and last updated on June 25, 2025.
313 long documents averaging 65.1 pages serve as an evaluation set for multimodal document information retrieval. The documents are categorized into ten domains including research reports, academic papers, financial reports, and government documents. The dataset, created by MMDocIR, features a modality distribution of 60.4% text, 18.8% image, and includes tables.
An open-source audio dataset for detecting infant cries and snoring sounds. It contains over 4 hours of audio data with corresponding annotations, created by QingyuLiu1 and last updated in June 2025.
Opens2S Datasets provide audio files for speech responses in English and Chinese. The dataset was authored by CASIA-LM and was last updated on July 8, 2025. The description indicates the data is split into compressed parts for English ('en_response_wav') and Chinese ('zh_response_wav') languages.
A dataset of creative text responses generated by the GPT-4.5, o3-mini, and DeepSeek-R1 language models. It contains responses to prompts for tasks like writing movie reviews and short stories, with styles enhanced by custom creative rubrics. The dataset was created by author 'vicgalle' and was last updated on June 16, 2025.
Highlighting Tswana language news articles sourced from the Daily News Dikgang, specifically formatted for the Massive Text Embedding Benchmark (MTEB). It provides labeled examples within the written news domain to support text-to-classification (t2c) tasks for low-resource language evaluation.
Korean memes, neologisms, and cultural phenomena are described in this dataset to enhance large language models' understanding and generation of Korean. It was created by Hugging Face KREW and last updated on June 5, 2025. The dataset provides detailed explanations of terms, including their definitions, origins, usage, and cultural context.
Wikimedia's Structured Wikipedia dataset is an early beta release containing all articles from the English and French language editions of Wikipedia. Articles are pre-parsed into structured JSON files with a consistent schema, including infoboxes. The dataset was last updated on 2025-05-29.
ViC-Bench is a dataset for benchmarking Multimodal Large Language Models (MLLMs) on visual reasoning tasks. It was created by meituan-longcat and last updated on May 30, 2025. The dataset likely contains tasks designed to evaluate models using Visual-Interleaved Chain-of-Thought reasoning.
788,984 peer reviews for submissions to the ICLR and NeurIPS conferences over an 8-year period. This dataset, created by IntelLabs, pairs human-written reviews with AI-generated counterparts produced by five large language models, including GPT-4o and Claude Sonnet. It was last updated on May 27, 2025.
News articles scraped from over 20 international publishers that release content under CC BY or CC BY-SA licenses. The dataset is hosted by common-pile on Hugging Face and was last updated on June 6, 2025. Sources include outlets such as 360info, Africa is a Country, Alt News, Global Voices, and Meduza.
SciNews is a dataset for generating scientific news from scholarly articles, created by dongqi-me and last updated in June 2025. It supports text generation, summarization, and classification tasks. The dataset contains text in English and is tagged with a US region focus.
A text corpus derived from the Chinese-language Wikipedia dump dated 2025-06-01. Articles were extracted from the original XML archive using Gensim and converted to Markdown format via post-processing. The dataset was uploaded by author Blaze7451 on 2025-06-25.
Over 16 million pages of historic US newspapers annotated with bounding boxes, predicted visual types, and OCR content. This dataset is a Parquet-converted version of the Newspaper Navigator dataset from the Library of Congress, originally released as JSON. The work was carried out as part of a project by Benjamin Germain Lee et al.
JMMMU is a Japanese multimodal benchmark designed to evaluate the performance of large multimodal models. The dataset was created by analyzing the existing MMMU benchmark for cultural dependencies and adapting culture-agnostic subjects with native Japanese speakers. The dataset was authored by JMMMU and last updated on the Hugging Face platform in May 2025.
Over 100,000 synthetic passport photos from more than 100 countries, created by UniDataPro. This dataset is intended for computer vision research and development, allowing work on passport verification and biometric tasks without using real personal data. The dataset was last updated on May 22, -05-22 08:33:39.
A collection of television scripts from the 'Black Mirror' anthology series, extracted from web sources using Jina scraping tools. The data is organized into a DatasetDict format, providing structured text for natural language processing and creative writing tasks focused on speculative fiction.
Dutch Book Review Sentiment contains text data for sentiment analysis tasks. The dataset was created by mteb and was last updated in July 2025. It is hosted on HuggingFace with platform tags indicating it is part of the Massive Text Embedding Benchmark (MTEB) and contains text modality.