The open-source xlm-roberta-large-xnli-anli model - Free deployment to facilitate zero-shot classification tasks

Xlm Roberta Large Xnli Anli

Developed by vicgalle

Based on the XLM-RoBERTa-large model, fine-tuned on multiple natural language inference datasets, and can be directly used for zero-shot classification tasks.

Large Language Model

Transformers

OtherOpen Source License:MIT #Multilingual zero-shot classification #Natural language inference #High-precision cross-lingual

Downloads 8,267

Release Time : 3/2/2022

Model Overview

This model is a multilingual natural language inference model that supports zero-shot classification tasks, suitable for text classification needs in various languages.

Model Features

Multilingual support

Supports text classification in multiple languages, including Spanish, French, etc.

Zero-shot classification

Can be directly used for classification tasks without fine-tuning, supporting custom labels.

High performance

Performs excellently on XNLI and ANLI datasets, with accuracy exceeding 93% especially on Spanish and French XNLI.

Model Capabilities

Zero-shot classification

Multilingual text understanding

Natural language inference

Use Cases

Text classification

Multilingual sentiment analysis

Use this model for sentiment classification of texts in multiple languages.

High-accuracy sentiment classification results.

Topic classification

Classify texts into custom topic labels.

Example shows a classification accuracy as high as 99.9% for the 'travel' label.

🚀 XLM-RoBERTa-large-XNLI-ANLI

The XLM-RoBERTa-large model is fine-tuned on multiple NLI datasets and is ready for zero-shot classification, offering high accuracy across various languages and tasks.

🚀 Quick Start

The model can be loaded using the zero-shot-classification pipeline as follows:

from transformers import pipeline
classifier = pipeline("zero-shot-classification", 
                       model="vicgalle/xlm-roberta-large-xnli-anli")

You can then use this pipeline to classify sequences into any of the class names you specify:

sequence_to_classify = "Algún día iré a ver el mundo"
candidate_labels = ['viaje', 'cocina', 'danza']
classifier(sequence_to_classify, candidate_labels)
#{'sequence': 'Algún día iré a ver el mundo',
#'labels': ['viaje', 'danza', 'cocina'],
#'scores': [0.9991760849952698, 0.0004178212257102132, 0.0004059972707182169]}

✨ Features

Multilingual Support: Suitable for zero-shot classification in multiple languages.
Fine-tuned on NLI Datasets: Trained on datasets such as MNLI, XNLI, and facebook/anli to enhance performance.

📦 Installation

There is no specific installation steps provided in the original README. If you want to use this model, you need to have the transformers library installed. You can install it using the following command:

pip install transformers

📚 Documentation

Accuracy on Test Datasets

Here are the accuracies for several test datasets:

	XNLI-es	XNLI-fr	ANLI-R1	ANLI-R2	ANLI-R3
xlm-roberta-large-xnli-anli	93.7%	93.2%	68.5%	53.6%	49.0%

Widget Examples

The model comes with some widget examples for quick testing:

Example 1:
- Text: De pugna erat fantastic. Nam Crixo decem quam dilexit et praeciderunt caput aemulus.
- Candidate Labels: violent, peaceful
Example 2:
- Text: La película empezaba bien pero terminó siendo un desastre.
- Candidate Labels: positivo, negativo, neutral
Example 3:
- Text: La película empezó siendo un desastre pero en general fue bien.
- Candidate Labels: positivo, negativo, neutral
Example 4:
- Text: ¿A quién vas a votar en 2020?
- Candidate Labels: Europa, elecciones, política, ciencia, deportes

📄 License

This project is licensed under the MIT license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご