---

XNLI: 18-Langauge NLI Dataset

Unlocking Multi-Language Natural Language Inference

By Huggingface Hub [source]

---

About this dataset

> The XNLI: Cross-Lingual Natural Language Inference Dataset is an 18-languages dataset containing information on natural language inference. This dataset has been designed to give researchers the ability to better understand the complexities of cross-lingual understanding by providing groups of premise, hypothesis, and label data in diverse languages. With this data, machine learning models can be trained and tested in both English and various non-English languages - such as Spanish, Arabic, Russian - for performance optimization in AI applications. Each entry of this dataset contains unique premise sentences as well as an associated hypothesis statement which incorporates a label (either entailment, neutral or contradiction) regarding the implication held between them. So whether your focus is language modeling or natural language processing, the XNLI dataset offers a wealth of study material that can open up new research opportunities for you!

More Datasets

> For more datasets, click here.

Featured Notebooks

> - 🚨 Your notebook can be here! 🚨!

How to use the dataset

> This dataset, XNLI: Cross-Lingual Natural Language Inference, offers an interesting opportunity to benchmark models in the field of natural language processing. It contains parallel inference examples in multiple languages for testing and validating natural language inference. This guide will provide an overview of the data and instructions on how to use it. > > The XNLI dataset consists of three sub-datasets: en_test.csv, el_validation.csv, and ur_test.csv. Each csv file consists of three columns: premise, hypothesis, and label. The premise column provides a statement or phrase; the hypothesis column presents a new statement that could be true or false according to the statement in the premise; finally, the label column indicates whether this sentence is an entailment (1), contradiction (-1) or neither (0). > > To get started with XNLI using machine learning algorithms such as Recurrent Neural Networks (RNNs) or Long short-term memory networks (LSTMs), you can use one of two methods: multi-language training with translation transfer learning or single language training with mono-lingual datasets like en_test and ur_test for English and Urdu respectively as your target language datasets for training models on NLI tasks . Depending on which setup you choose to go with these datasets can both be used to produce powerful modern natural language understanding systems as well as patient diagnosis assistance tools utilizing NLI capabilities such as those trained on healthcare domain data like medical chit chat conversations etc., > > To start building a model using multi-language transfer learning first split up your classes into two separate sets that will make up your training dataset which are then immediately used during validation through utilising cross validation techniques like kfold methods when running all experiments necessary for hyperparemeter tuning procedures while concurrently working on transforming text from all languages into desired target english statements by employing /name dropped/ translation services API's either directly within code application by making call outs stating .^ specific parameters configured inside development environment variabels before then performing model fitting procedures straight away but Bear in mind certain points especially when fine tuning any deeplearning architecture because let’s not forget that since same eggs were thrown once during early stages till now so same principles might apply again here too! > > For starting with building Mono lingual NLP systems based on this xnli datasets such process may look something like initially loading up text data from individual files from chosen destination folder followed by cleaning it accordingly fitting token

Research Ideas

> - Training and testing a cross-lingual NLI model for language translation applications. > - Building a sentiment analyzer that can accurately classify sentiment in 18 different languages. > - Constructing an AI assistant that is capable of understanding natural language in 18 languages and providing appropriate responses accordingly

Acknowledgements

> If you use this dataset in your research, please credit the original authors. > Data Source > >

License

> > > License: CC0 1.0 Universal (CC0 1.0) - Public Domain Dedication > No Copyright - You can copy, modify, distribute and perform the work, even for commercial purposes, all without asking permission. See Other Information.

Columns

File: el_validation.csv

Column name	Description
premise	The premise of the natural language inference. (String)
hypothesis	The hypothesis of the natural language inference. (String)
label	The label of the natural language inference, such as entailment, neutral or contradiction. (String)

---

File: en_test.csv

Column name	Description
premise	The premise of the natural language inference. (String)
hypothesis	The hypothesis of the natural language inference. (String)
label	The label of the natural language inference, such as entailment, neutral or contradiction. (String)

---

File: ur_test.csv

Column name	Description
premise	The premise of the natural language inference. (String)
hypothesis	The hypothesis of the natural language inference. (String)
label	The label of the natural language inference, such as entailment, neutral or contradiction. (String)

Acknowledgements

> If you use this dataset in your research, please credit the original authors. > If you use this dataset in your research, please credit Huggingface Hub.