Language Modelling

The MNIST database (Modified National Institute of Standards and Technology database) is a large collection of handwritten digits. It has a training set of 60,000 examples, and a test set of 10,000 examples. It is a subset of a larger NIST Special Database 3 (digits written by employees of the United States Census Bureau) and Special Database 1 (digits written by high school students) which contain monochrome images of handwritten digits. The digits have been size-normalized and centered in a fixed-size image. The original black and white (bilevel) images from NIST were size normalized to fit in a 20x20 pixel box while preserving their aspect ratio. The resulting images contain grey levels as a result of the anti-aliasing technique used by the normalization algorithm. the images were centered in a 28x28 image by computing the center of mass of the pixels, and translating the image so as to position this point at the center of the 28x28 field. Source: http://yann.lecun.com/exdb/mnist/ Image Source: https://en.wikipedia.org/wiki/MNIST_database#/media/File:MnistExamples.png

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. Source: http://nlpprogress.com/english/sentiment_analysis.html Image Source: Maas et al

The PubMed dataset consists of 19717 scientific publications from PubMed database pertaining to diabetes classified into one of three classes. The citation network consists of 44338 links. Each publication in the dataset is described by a TF/IDF weighted word vector from a dictionary which consists of 500 unique words. Source: https://linqs.soe.ucsc.edu/data

The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike License. Compared to the preprocessed version of Penn Treebank (PTB), WikiText-2 is over 2 times larger and WikiText-103 is over 110 times larger. The WikiText dataset also features a far larger vocabulary and retains the original case, punctuation and numbers - all of which are removed in PTB. As it is composed of full articles, the dataset is well suited for models that can take advantage of long term dependencies. Source: The WikiText Long Term Dependency Language Modeling Dataset Image Source: https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/

The English Penn Treebank (PTB) corpus, and in particular the section of the corpus corresponding to the articles of Wall Street Journal (WSJ), is one of the most known and used corpus for the evaluation of models for sequence labelling. The task consists of annotating each word with its Part-of-Speech tag. In the most common split of this corpus, sections from 0 to 18 are used for training (38 219 sentences, 912 344 tokens), sections from 19 to 21 are used for validation (5 527 sentences, 131 768 tokens), and sections from 22 to 24 are used for testing (5 462 sentences, 129 654 tokens). The corpus is also commonly used for character-level and word-level Language Modelling. Source: Seq2Biseq: Bidirectional Output-wise Recurrent Neural Networks for Sequence Modelling Image Source: https://dl.acm.org/doi/10.5555/972470.972475

C4 is a colossal, cleaned version of Common Crawl's web crawl corpus. It was based on Common Crawl dataset: https://commoncrawl.org. It was used to train the T5 text-to-text Transformer models. The dataset can be downloaded in a pre-processed form from allennlp.

CoNLL-2003 is a named entity recognition dataset released as a part of CoNLL-2003 shared task: language-independent named entity recognition. The data consists of eight files covering two languages: English and German. For each of the languages there is a training file, a development file, a test file and a large file with unannotated data. The English data was taken from the Reuters Corpus. This corpus consists of Reuters news stories between August 1996 and August 1997. For the training and development set, ten days worth of data were taken from the files representing the end of August 1996. For the test set, the texts were from December 1996. The preprocessed raw data covers the month of September 1996. The text for the German data was taken from the ECI Multilingual Text Corpus. This corpus consists of texts in many languages. The portion of data that was used for this task, was extracted from the German newspaper Frankfurter Rundshau. All three of the training, development and test sets were taken from articles written in one week at the end of August 1992. The raw data were taken from the months of September to December 1992. | English data | Articles | Sentences | Tokens | LOC | MISC | ORG | PER | |-------------------|----------|-----------|---------|------|------|------|------| | Training set | 946 | 14,987 | 203,621 | 7140 | 3438 | 6321 | 6600 | | Development set | 216 | 3,466 | 51,362 | 1837 | 922 | 1341 | 1842 | | Test set | 231 | 3,684 | 46,435 | 1668 | 702 | 1661 | 1617 | Number of articles, sentences, tokens and entities (locations, miscellaneous, organizations, and persons) in English data files. | German data | Articles | Sentences | Tokens | LOC | MISC | ORG | PER | |-------------------|----------|-----------|---------|------|------|------|------| | Training set | 553 | 12,705 | 206,931 | 4363 | 2288 | 2427 | 2773 | | Development set | 201 | 3,068 | 51,444 | 1181 | 1010 | 1241 | 1401 | | Test set | 155 | 3,160 | 51,943 | 1035 | 670 | 773 | 1195 | Number of articles, sentences, tokens and entities (locations, miscellaneous, organizations, and persons) in German data files.

The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike License. Compared to the preprocessed version of Penn Treebank (PTB), WikiText-2 is over 2 times larger and WikiText-103 is over 110 times larger. The WikiText dataset also features a far larger vocabulary and retains the original case, punctuation and numbers - all of which are removed in PTB. As it is composed of full articles, the dataset is well suited for models that can take advantage of long term dependencies. Source: The WikiText Long Term Dependency Language Modeling Dataset Image Source: https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/

The Pile is a 825 GiB diverse, open source language modelling data set that consists of 22 smaller, high-quality datasets combined together. Datasheet: Datasheet for the Pile

The Beyond the Imitation Game Benchmark (BIG-bench) is a collaborative benchmark intended to probe large language models and extrapolate their future capabilities. Big-bench include more than 200 tasks. Image source: https://arxiv.org/pdf/2206.04615.pdf

The Sentences Involving Compositional Knowledge (SICK) dataset is a dataset for compositional distributional semantics. It includes a large number of sentence pairs that are rich in the lexical, syntactic and semantic phenomena. Each pair of sentences is annotated in two dimensions: relatedness and entailment. The relatedness score ranges from 1 to 5, and Pearson’s r is used for evaluation; the entailment relation is categorical, consisting of entailment, contradiction, and neutral. There are 4439 pairs in the train split, 495 in the trial split used for development and 4906 in the test split. The sentence pairs are generated from image and video caption datasets before being paired up using some algorithm. Source: Multi-Label Transfer Learning for Multi-Relational Semantic Similarity Image Source: https://www.researchgate.net/figure/Example-of-SICK-dataset-sentence-expansion-process-14_fig1_344863619

BookCorpus is a large collection of free novel books written by unpublished authors, which contains 11,038 books (around 74M sentences and 1G words) of 16 different sub-genres (e.g., Romance, Historical, Adventure, etc.). Source: Temporal Event Knowledge Acquisition via Identifying Narratives

The LAMBADA (LAnguage Modeling Broadened to Account for Discourse Aspects) benchmark is an open-ended cloze task which consists of about 10,000 passages from BooksCorpus where a missing target word is predicted in the last sentence of each passage. The missing word is constrained to always be the last word of the last sentence and there are no candidate words to choose from. Examples were filtered by humans to ensure they were possible to guess given the context, i.e., the sentences in the passage leading up to the last sentence. Examples were further filtered to ensure that missing words could not be guessed without the context, ensuring that models attempting the dataset would need to reason over the entire paragraph to answer questions. Source: Recent Advances in Natural Language Inference:A Survey of Benchmarks, Resources, and Approaches Image Source: https://arxiv.org/pdf/1606.06031.pdf

The task of PubMedQA is to answer research questions with yes/no/maybe (e.g.: Do preoperative statins reduce atrial fibrillation after coronary artery bypass grafting?) using the corresponding abstracts. PubMedQA has 1k expert labeled, 61.2k unlabeled and 211.3k artificially generated QA instances. Source: PubMedQA Image Source: https://arxiv.org/pdf/1909.06146v1.pdf

LAnguage Model Analysis (LAMA) consists of a set of knowledge sources, each comprised of a set of facts. LAMA is a probe for analyzing the factual and commonsense knowledge contained in pretrained language models. Source: Language Models as Knowledge Bases?

OpenSubtitles is collection of multilingual parallel corpora. The dataset is compiled from a large database of movie and TV subtitles and includes a total of 1689 bitexts spanning 2.6 billion sentences across 60 languages.

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). Source: RoBERTa: A Robustly Optimized BERT Pretraining Approach

WiC is a benchmark for the evaluation of context-sensitive word embeddings. WiC is framed as a binary classification task. Each instance in WiC has a target word w, either a verb or a noun, for which two contexts are provided. Each of these contexts triggers a specific meaning of w. The task is to identify if the occurrences of w in the two contexts correspond to the same meaning or not. In fact, the dataset can also be viewed as an application of Word Sense Disambiguation in practise. Source: WiC: the Word-in-Context Dataset for Evaluating Context-Sensitive Meaning Representations

Clotho is an audio captioning dataset, consisting of 4981 audio samples, and each audio sample has five captions (a total of 24 905 captions). Audio samples are of 15 to 30 s duration and captions are eight to 20 words long. Source: https://zenodo.org/record/3490684 Image Source: https://arxiv.org/abs/1910.09387

AISHELL-1 is a corpus for speech recognition research and building speech recognition systems for Mandarin. Source: AISHELL-1: An Open-Source Mandarin Speech Corpus and A Speech Recognition Baseline

Libri-Light is a collection of spoken English audio suitable for training speech recognition systems under limited or no supervision. It is derived from open-source audio books from the LibriVox project. It contains over 60K hours of audio. Source: Libri-Light: A Benchmark for ASR with Limited or No Supervision

XQuAD (Cross-lingual Question Answering Dataset) is a benchmark dataset for evaluating cross-lingual question answering performance. The dataset consists of a subset of 240 paragraphs and 1190 question-answer pairs from the development set of SQuAD v1.1 (Rajpurkar et al., 2016) together with their professional translations into ten languages: Spanish, German, Greek, Russian, Turkish, Arabic, Vietnamese, Thai, Chinese, and Hindi. Consequently, the dataset is entirely parallel across 11 languages. Source: XQuAD Image Source: https://arxiv.org/pdf/1910.11856v3.pdf

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. Source: https://arxiv.org/pdf/1809.04444.pdf

Long-range arena (LRA) is an effort toward systematic evaluation of efficient transformer models. The project aims at establishing benchmark tasks/datasets using which we can evaluate transformer-based models in a systematic way, by assessing their generalization power, computational efficiency, memory foot-print, etc. Long-Range Arena is specifically focused on evaluating model quality under long-context scenarios. The benchmark is a suite of tasks consisting of sequences ranging from 1K to 16K tokens, encompassing a wide range of data types and modalities such as text, natural, synthetic images, and mathematical expressions requiring similarity, structural, and visual-spatial reasoning. Description from: Long Range Arena : A Benchmark for Efficient Transformers

PAWS-X contains 23,659 human translated PAWS evaluation pairs and 296,406 machine translated training pairs in six typologically distinct languages: French, Spanish, German, Chinese, Japanese, and Korean. All translated pairs are sourced from examples in PAWS-Wiki. Source: Google Research Image Source: https://arxiv.org/pdf/1908.11828v1.pdf

ATOMIC is an atlas of everyday commonsense reasoning, organized through 877k textual descriptions of inferential knowledge. Compared to existing resources that center around taxonomic knowledge, ATOMIC focuses on inferential knowledge organized as typed if-then relations with variables (e.g., "if X pays Y a compliment, then Y will likely return the compliment"). Source: ATOMIC: An Atlas of Machine Commonsense for If-Then Reasoning Image Source: https://homes.cs.washington.edu/~msap/atomic/

ELI5 is a dataset for long-form question answering. It contains 270K complex, diverse questions that require explanatory multi-sentence answers. Web search results are used as evidence documents to answer each question. ELI5 is also a task in Dodecadialogue. Source: ELI5 Image Source: https://arxiv.org/pdf/1907.09190v1.pdf

The OLID is a hierarchical dataset to identify the type and the target of offensive texts in social media. The dataset is collected on Twitter and publicly available. There are 14,100 tweets in total, in which 13,240 are in the training set, and 860 are in the test set. For each tweet, there are three levels of labels: (A) Offensive/Not-Offensive, (B) Targeted-Insult/Untargeted, (C) Individual/Group/Other. The relationship between them is hierarchical. If a tweet is offensive, it can have a target or no target. If it is offensive to a specific target, the target can be an individual, a group, or some other objects. This dataset is used in the OffensEval-2019 competition in SemEval-2019. Source: Kungfupanda at SemEval-2020 Task 12: BERT-Based Multi-Task Learning for Offensive Language Detection Image Source: https://arxiv.org/pdf/1902.09666.pdf

A new open-vocabulary language modelling benchmark derived from books. Source: Compressive Transformers for Long-Range Sequence Modelling

The One Billion Word dataset is a dataset for language modeling. The training/held-out data was produced from the WMT 2011 News Crawl data using a combination of Bash shell and Perl scripts.

The BLUE benchmark consists of five different biomedicine text-mining tasks with ten corpora. These tasks cover a diverse range of text genres (biomedical literature and clinical notes), dataset sizes, and degrees of difficulty and, more importantly, highlight common biomedicine text-mining challenges. Source: Transfer Learning in Biomedical Natural Language Processing: An Evaluation of BERT and ELMo on Ten Benchmarking Datasets

BLiMP is a challenge set for evaluating what language models (LMs) know about major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each containing 1000 minimal pairs isolating specific contrasts in syntax, morphology, or semantics. The data is automatically generated according to expert-crafted grammars. Aggregate human agreement with the labels is 96.4%. Source: BLiMP Image Source: https://arxiv.org/pdf/1912.00582v3.pdf

Multi-News, consists of news articles and human-written summaries of these articles from the site newser.com. Each summary is professionally written by editors and includes links to the original articles cited. Source: Multi-News: a Large-Scale Multi-Document Summarization Dataset and Abstractive Hierarchical Model Image Source: https://arxiv.org/pdf/1906.01749.pdf

Social Interaction QA (SIQA) is a question-answering benchmark for testing social commonsense intelligence. Contrary to many prior benchmarks that focus on physical or taxonomic knowledge, Social IQa focuses on reasoning about people’s actions and their social implications. For example, given an action like "Jesse saw a concert" and a question like "Why did Jesse do this?", humans can easily infer that Jesse wanted "to see their favorite performer" or "to enjoy the music", and not "to see what's happening inside" or "to see if it works". The actions in Social IQa span a wide variety of social situations, and answer candidates contain both human-curated answers and adversarially-filtered machine-generated candidates. Social IQa contains over 37,000 QA pairs for evaluating models’ abilities to reason about the social implications of everyday events and situations. Source: Social IQA Image Source: https://arxiv.org/pdf/1904.09728.pdf

WritingPrompts is a large dataset of 300K human-written stories paired with writing prompts from an online forum. Source: Hierarchical Neural Story Generation

A dataset of large scale alignments between Wikipedia abstracts and Wikidata triples. T-REx consists of 11 million triples aligned with 3.09 million Wikipedia abstracts (6.2 million sentences). Source: T-REx: A Large Scale Alignment of Natural Language with Knowledge Base Triples