214 datasets found

CBBQ is a Chinese Bias Benchmark dataset curated with Human-AI Collaboration for Large Language Models. It consists of over 100K questions jointly constructed by human experts...

A benchmark for question answering research is introduced, which includes a large dataset of natural questions.

A natural language fmri dataset for voxelwise encoding models.

The MSVD dataset is a collection of 1970 open domain clips from YouTube, annotated with variable-length captions.

A dataset of annotations for the Interactive Gameplay dataset.

A dataset of natural language commands written by crowd-sourced workers for an interactive Minecraft game.

A dataset of natural language commands written by crowd-sourced workers for an interactive Minecraft game.

AI-hub dialogue dataset for Korean dialogue processing

KLUE benchmark dataset for Korean language understanding

Syntactically Incomplete Korean (SIKO) dataset for Korean language models

A dataset of 297,845 preprocessed Python source files for training and evaluation of natural language-guided programming (NLGP) models.

A dataset of 201 curated test cases for natural language-guided programming (NLGP) to evaluate the quality of code predictions.

The dataset contains 2.4 million reviews from five different MOOC platforms: Udemy, Coursera, Domestika, Platzi, and Crehana.

The Integrated Crisis Early Warning System (ICEWS) dataset is a real-time stream of news stories ingested and processed to create a final dataset of events.

ROC-Stories: A Corpus for Evaluating Story Generation Models

NGEP: A Graph-based Event Planning Framework for Story Generation

RealityTalk is a system that augments live video presentations with speech-driven interactive virtual elements.

The datasets developed for the SIGMORPHON 2019 lemmatization task are annotated according to the Unimorph schema guidelines.

The dataset used for the automatic text summarization task, containing documents in three languages.

The dataset is a collection of online videos from YouTube, with comments and metadata. It is used to evaluate the performance of the Online Video Clickbait Protector (OVCP) scheme.