20,499 datasets found

The Building Educational Applications (BEA) shared task on GEC provides datasets including the Cambridge English Write & Improve corpus, which is composed of texts written...

The CoNLL 2014 shared task dataset is comprised of essays written by undergraduate students, annotated for grammatical errors.

The First Certificate in English (FCE) dataset contains essays written by non-native learners of English assessed in a language exam, annotated for language errors and...

The dataset consists of parallel data from various corpuses used for training and evaluating the bilingual BERT model for translation quality estimation.

QuAC is a dataset for question answering in a conversational context, requiring understanding of the multi-turn dialogue history to provide contextually relevant answers derived...

A dataset of 3,000 sarcastic tweets, each interpreted by five human judges, focusing on the task of sarcasm interpretation.

The dataset contains 4,762 pairs of sarcastic messages and hearer interpretations, collected through a crowdsourcing experiment.

The MedNLI dataset is used to predict the entailment relation between a pair of sentences, with premises taken from doctors' notes in the clinical dataset MIMIC-III.

The MultiNLI corpus is a dataset designed to assist in learning natural language inference, featuring sentence pairs labeled as entailment, neutral, or contradiction, which aid...

The dataset comprises 13023 accounts of sexism, including first-person accounts from survivors, each tagged with at least one of 23 categories of sexism.

The ConvAI2 dataset, derived from Persona-Chat, contains dialogues between crowdworkers who role-play as assigned personas, enabling the development of conversational agents...

The REST dataset is derived from restaurant reviews, also containing review sentences and aspect sentiment annotations for aspect-based sentiment analysis.

The LAPTOP dataset is used for aspect-based sentiment analysis, containing review sentences along with gold standard aspect sentiment annotations.

OCNLI is a dataset for natural language inference adapted for Chinese language, consisting of premise-hypothesis pairs.

BQ Corpus is a large-scale dataset for sentence semantic equivalence identification in Chinese.

LCQMC is a large-scale Chinese question matching corpus used for determining the semantic equivalence of question pairs.

TNEWS is a short text classification dataset consisting of news titles and keywords requiring classification into one of 15 classes.

THUCNews is a dataset used for news categorization tasks in different genres, containing 50K news articles in ten domains.

ChnSentiCorp is a dataset used for sentiment classification in Chinese documents, where the text is classified into positive or negative labels.

CJRC is a dataset for machine reading comprehension specializing in Chinese legal judgments, containing yes/no questions, no-answer questions, and span-extraction questions.