General Language Understanding Evaluation (GLUE) benchmark is a collection of nine natural language understanding tasks, including single-sentence tasks CoLA and SST-2, similarity and paraphrasing tasks MRPC, STS-B and QQP, and natural language inference tasks MNLI, QNLI, RTE and WNLI.
2,731 PAPERS • 25 BENCHMARKS
The Stanford Sentiment Treebank is a corpus with fully labeled parse trees that allows for a complete analysis of the compositional effects of sentiment in language. The corpus is based on the dataset introduced by Pang and Lee (2005) and consists of 11,855 single sentences extracted from movie reviews. It was parsed with the Stanford parser and includes a total of 215,154 unique phrases from those parse trees, each annotated by 3 human judges.
2,032 PAPERS • 9 BENCHMARKS
The IMDb Movie Reviews dataset is a binary sentiment analysis dataset consisting of 50,000 reviews from the Internet Movie Database (IMDb) labeled as positive or negative. The dataset contains an even number of positive and negative reviews. Only highly polarizing reviews are considered. A negative review has a score ≤ 4 out of 10, and a positive review has a score ≥ 7 out of 10. No more than 30 reviews are included per movie. The dataset contains additional unlabeled data.
1,586 PAPERS • 11 BENCHMARKS
AG News (AG’s News Corpus) is a subdataset of AG's corpus of news articles constructed by assembling titles and description fields of articles from the 4 largest classes (“World”, “Sports”, “Business”, “Sci/Tech”) of AG’s Corpus. The AG News contains 30,000 training and 1,900 test samples per class.
780 PAPERS • 10 BENCHMARKS
DBpedia (from "DB" for "database") is a project aiming to extract structured content from the information created in the Wikipedia project. DBpedia allows users to semantically query relationships and properties of Wikipedia resources, including links to other related datasets.
554 PAPERS • 4 BENCHMARKS
The RCV1 dataset is a benchmark dataset on text categorization. It is a collection of newswire articles producd by Reuters in 1996-1997. It contains 804,414 manually labeled newswire documents, and categorized with respect to three controlled vocabularies: industries, topics and regions.
321 PAPERS • 6 BENCHMARKS
Dataset of hate speech annotated on Internet forum posts in English at sentence-level. The source forum in Stormfront, a large online community of white nacionalists. A total of 10,568 sentence have been been extracted from Stormfront and classified as conveying hate speech or not.
162 PAPERS • 1 BENCHMARK
OpenWebText is an open-source recreation of the WebText corpus. The text is web content extracted from URLs shared on Reddit with at least three upvotes. (38GB).
133 PAPERS • NO BENCHMARKS YET
e-SNLI is used for various goals, such as obtaining full sentence justifications of a model's decisions, improving universal sentence representations and transferring to out-of-domain NLI datasets.
124 PAPERS • 1 BENCHMARK
The Yahoo! Answers topic classification dataset is constructed using 10 largest main categories. Each class contains 140,000 training samples and 6,000 testing samples. Therefore, the total number of training samples is 1,400,000 and testing samples 60,000 in this dataset. From all the answers and other meta-information, we only used the best answer content and the main category information. Source:github
121 PAPERS • 2 BENCHMARKS
Dataset composed of online banking queries annotated with their corresponding intents.
100 PAPERS • 5 BENCHMARKS
CARER is an emotion dataset collected through noisy labels, annotated via distant supervision as in (Go et al., 2009).
97 PAPERS • 4 BENCHMARKS
CLUE is a Chinese Language Understanding Evaluation benchmark. It consists of different NLU datasets. It is a community-driven project that brings together 9 tasks spanning several well-established single-sentence/sentence-pair classification tasks, as well as machine reading comprehension, all on original Chinese text.
95 PAPERS • 8 BENCHMARKS
This dataset is for evaluating the performance of intent classification systems in the presence of "out-of-scope" queries, i.e., queries that do not fall into any of the system-supported intent classes. The dataset includes both in-scope and out-of-scope data.
73 PAPERS • 5 BENCHMARKS
Text Retrieval Conference Question Answering (TrecQA) is a dataset created from the TREC-8 (1999) to TREC-13 (2004) Question Answering tracks. There are two versions of TrecQA: raw and clean. Both versions have the same training set but their development and test sets differ. The commonly used clean version of the dataset excludes questions in development and test sets with no answers or only positive/negative answers. The clean version has 1,229/65/68 questions and 53,417/1,117/1,442 question-answer pairs for the train/dev/test split.
72 PAPERS • 3 BENCHMARKS
TweetEval introduces an evaluation framework consisting of seven heterogeneous Twitter-specific classification tasks.
72 PAPERS • 2 BENCHMARKS
MTEB is a benchmark that spans 8 embedding tasks covering a total of 56 datasets and 112 languages. The 8 task types are Bitext mining, Classification, Clustering, Pair Classification, Reranking, Retrieval, Semantic Textual Similarity and Summarisation. The 56 datasets contain varying text lengths and they are grouped into three categories: Sentence to sentence, Paragraph to paragraph, and Sentence to paragraph.
53 PAPERS • 8 BENCHMARKS
Web of Science (WOS) is a document classification dataset that contains 46,985 documents with 134 categories which include 7 parents categories.
48 PAPERS • 4 BENCHMARKS
This dataset contains product reviews and metadata from Amazon, including 142.8 million reviews spanning May 1996 - July 2014.
33 PAPERS • 6 BENCHMARKS
The Yelp Reviews Polarity dataset is obtained from the Yelp Dataset Challenge in 2015 (1,569,264 samples that have review text).
33 PAPERS • 1 BENCHMARK
PearRead is a dataset of scientific peer reviews. The dataset consists of over 14K paper drafts and the corresponding accept/reject decisions in top-tier venues including ACL, NIPS and ICLR, as well as over 10K textual peer reviews written by experts for a subset of the papers.
32 PAPERS • NO BENCHMARKS YET
BLURB is a collection of resources for biomedical natural language processing. In general domains such as newswire and the Web, comprehensive benchmarks and leaderboards such as GLUE have greatly accelerated progress in open-domain NLP. In biomedicine, however, such resources are ostensibly scarce. In the past, there have been a plethora of shared tasks in biomedical NLP, such as BioCreative, BioNLP Shared Tasks, SemEval, and BioASQ, to name just a few. These efforts have played a significant role in fueling interest and progress by the research community, but they typically focus on individual tasks. The advent of neural language models such as BERTs provides a unifying foundation to leverage transfer learning from unlabeled text to support a wide range of NLP applications. To accelerate progress in biomedical pretraining strategies and task-specific methods, it is thus imperative to create a broad-coverage benchmark encompassing diverse biomedical tasks.
31 PAPERS • 2 BENCHMARKS
MR Movie Reviews is a dataset for use in sentiment-analysis experiments. Available are collections of movie-review documents labeled with respect to their overall sentiment polarity (positive or negative) or subjective rating (e.g., "two and a half stars") and sentences labeled with respect to their subjectivity status (subjective or objective) or polarity.
27 PAPERS • 3 BENCHMARKS
The 20 Newsgroups data set is a collection of approximately 20,000 newsgroup documents, partitioned (nearly) evenly across 20 different newsgroups.
26 PAPERS • 6 BENCHMARKS
Evidence Inference is a corpus for this task comprising 10,000+ prompts coupled with full-text articles describing RCTs.
26 PAPERS • NO BENCHMARKS YET
EURLEX57K is a new publicly available legal LMTC dataset, dubbed EURLEX57K, containing 57k English EU legislative documents from the EUR-LEX portal, tagged with ∼4.3k labels (concepts) from the European Vocabulary (EUROVOC).
21 PAPERS • NO BENCHMARKS YET
Korean Language Understanding Evaluation (KLUE) benchmark is a series of datasets to evaluate natural language understanding capability of Korean language models. KLUE consists of 8 diverse and representative tasks, which are accessible to anyone without any restrictions. With ethical considerations in mind, we deliberately design annotation guidelines to obtain unambiguous annotations for all datasets. Furthermore, we build an evaluation system and carefully choose evaluations metrics for every task, thus establishing fair comparison across Korean language models.
19 PAPERS • 1 BENCHMARK
Moral Stories is a crowd-sourced dataset of structured narratives that describe normative and norm-divergent actions taken by individuals to accomplish certain intentions in concrete situations, and their respective consequences.
19 PAPERS • NO BENCHMARKS YET
PubMed 200k RCT is new dataset based on PubMed for sequential sentence classification. The dataset consists of approximately 200,000 abstracts of randomized controlled trials, totaling 2.3 million sentences. Each sentence of each abstract is labeled with their role in the abstract using one of the following classes: background, objective, method, result, or conclusion. The purpose of releasing this dataset is twofold. First, the majority of datasets for sequential short-text classification (i.e., classification of short texts that appear in sequences) are small: the authors hope that releasing a new large dataset will help develop more accurate algorithms for this task. Second, from an application perspective, researchers need better tools to efficiently skim through the literature. Automatically classifying each sentence in an abstract would help researchers read abstracts more efficiently, especially in fields where abstracts may be long, such as the medical field.
18 PAPERS • NO BENCHMARKS YET
A novel large dataset of social media posts from users with one or multiple mental health conditions along with matched control users.
16 PAPERS • NO BENCHMARKS YET
The Terms of Service dataset is a law dataset corresponding to the task of identifying whether contractual terms are potentially unfair. This is a binary classification task, where positive examples are potentially unfair contractual terms (clauses) from the terms of service in consumer contracts. Article 3 of the Directive 93/13 on Unfair Terms in Consumer Contracts defines an unfair contractual term as follows. A contractual term is unfair if: (1) it has not been individually negotiated; and (2) contrary to the requirement of good faith, it causes a significant imbalance in the parties rights and obligations, to the detriment of the consumer. The Terms of Service dataset consists of 9,414 examples.
16 PAPERS • 1 BENCHMARK
BioRED is a first-of-its-kind biomedical relation extraction dataset with multiple entity types (e.g. gene/protein, disease, chemical) and relation pairs (e.g. gene–disease; chemical–chemical) at the document level, on a set of600 PubMed abstracts. Furthermore, BioRED label each relation as describing either a novel finding or previously known background knowledge, enabling automated algorithms to differentiate between novel and background information.
14 PAPERS • 3 BENCHMARKS
FLUE is a French Language Understanding Evaluation benchmark. It consists of 5 tasks: Text Classification, Paraphrasing, Natural Language Inference, Constituency Parsing and Part-of-Speech Tagging, and Word Sense Disambiguation.
12 PAPERS • NO BENCHMARKS YET
Ohsumed includes medical abstracts from the MeSH categories of the year 1991. In [Joachims, 1997] were used the first 20,000 documents divided in 10,000 for training and 10,000 for testing. The specific task was to categorize the 23 cardiovascular diseases categories. After selecting the such category subset, the unique abstract number becomes 13,929 (6,286 for training and 7,643 for testing). As current computers can easily manage larger number of documents we make available all 34,389 cardiovascular diseases abstracts out of 50,216 medical abstracts contained in the year 1991.
11 PAPERS • 2 BENCHMARKS
The Overruling dataset is a law dataset corresponding to the task of determining when a sentence is overruling a prior decision. This is a binary classification task, where positive examples are overruling sentences and negative examples are non-overruling sentences extracted from legal opinions. In law, an overruling sentence is a statement that nullifies a previous case decision as a precedent, by a constitutionally valid statute or a decision by the same or higher ranking court which establishes a different rule on the point of law involved. The Overruling dataset consists of 2,400 sentences.
10 PAPERS • 1 BENCHMARK
OMICS is an extensive collection of knowledge for indoor service robots gathered from internet users. Currently, it contains 48 tables capturing different sorts of knowledge. Each tuple of the Help table maps a user desire to a task that may meet the desire (e.g., ⟨ “feel thirsty”, “by offering drink” ⟩). Each tuple of the Tasks/Steps table decomposes a task into several steps (e.g., ⟨ “serve a drink”, 0. “get a glass”, 1. “get a bottle”, 2. “fill class from bottle”, 3. “give class to person” ⟩). Given this, OMICS offers useful knowledge about hierarchism of naturalistic instructions, where a high-level user request (e.g., “serve a drink”) can be reduced to lower-level tasks (e.g., “get a glass”, ⋯). Another feature of OMICS is that elements of any tuple in an OMICS table are semantically related according to a predefined template. This facilitates the semantic interpretation of the OMICS tuples.
6 PAPERS • NO BENCHMARKS YET
A question type classification dataset with 6 classes for questions about a person, location, numeric information, etc. The test split has 500 questions, and the training split has 5452 questions.
6 PAPERS • 2 BENCHMARKS
Taiga is a corpus, where text sources and their meta-information are collected according to popular ML tasks.
5 PAPERS • NO BENCHMARKS YET
MuMiN is a misinformation graph dataset containing rich social media data (tweets, replies, users, images, articles, hashtags), spanning 21 million tweets belonging to 26 thousand Twitter threads, each of which have been semantically linked to 13 thousand fact-checked claims across dozens of topics, events and domains, in 41 different languages, spanning more than a decade.
4 PAPERS • 3 BENCHMARKS
This is an entity-level Twitter Sentiment Analysis dataset. For each message, the task is to judge the sentiment of the entire sentence towards a given entity. For example, A outperforms B is positive for entity A but negative for entity B. The dataset contains ~70K labeled training messages and 1K labeled validation messages. It is available online for free on Kaggle.
4 PAPERS • 1 BENCHMARK
Benchmark dataset for abstracts and titles of 100,000 ArXiv scientific papers. This dataset contains 10 classes and is balanced (exactly 10,000 per class). The classes include subcategories of computer science, physics, and math.
FreSaDa is a French satire dataset for cross-domain satire detection, which is composed of 11,570 articles from the news domain. The dataset samples have been split into training, validation and test, such that the training publication sources are distinct from the validation and test publication sources. This gives rise to a cross-domain (cross-source) satire detection task.
3 PAPERS • NO BENCHMARKS YET
Multilabeled News Dataset (MN-DS) is a dataset for news classification. It consists of 10,917 articles in 17 first-level and 109 second-level categories from 215 media sources.
The Medical Abstracts dataset contains 14,438 medical abstracts describing 5 different classes of patient conditions, with all of the dataset being annotated. The dataset is split into training and test sets.
3 PAPERS • 1 BENCHMARK
MuLD (Multitask Long Document Benchmark) is a set of 6 NLP tasks where the inputs consist of at least 10,000 words. The benchmark covers a wide variety of task types including translation, summarization, question answering, and classification. Additionally there is a range of output lengths from a single word classification label all the way up to an output longer than the input text.
3 PAPERS • 6 BENCHMARKS
SV-Ident comprises 4,248 sentences from social science publications in English and German. The data is the official data for the Shared Task: “Survey Variable Identification in Social Science Publications” (SV-Ident) 2022. Sentences are labeled with variables that are mentioned either explicitly or implicitly.
3 PAPERS • 2 BENCHMARKS
Text Classification Attack Benchmark (TCAB) is a dataset for analyzing, understanding, detecting, and labeling adversarial attacks against text classifiers. TCAB includes 1.5 million attack instances, generated by twelve adversarial attack targeting three classifiers trained on six source datasets for sentiment analysis and abuse detection in English. The process of generating attacks is automated, so that TCAB can easily be extended to incorporate new text attacks and better classifiers as they are developed.
Wikipedia Title is a dataset for learning character-level compositionality from the character visual characteristics. It consists of a collection of Wikipedia titles in Chinese, Japanese or Korean labelled with the category to which the article belongs.