XLM-RoBERTa Base FT UDPOS28 - CA Open - source Multilingual Part - of - Speech Tagging Model - Supports Multiple Languages Including Catalan

Home

Xlm Roberta Base Ft Udpos28 Ca

Developed by wietsedv

Multilingual POS tagging model based on XLM-RoBERTa, supporting Catalan and other languages

Sequence Labeling

Transformers

OtherOpen Source License:Apache-2.0 #Multilingual POS tagging #High accuracy #Catalan optimization

Downloads 24

Release Time : 3/2/2022

Model Overview

This model is a multilingual POS tagging model based on the XLM-RoBERTa architecture, specifically optimized for Catalan, capable of accurately identifying the part-of-speech categories of words in text.

Model Features

Multilingual support

Based on the XLM-RoBERTa architecture, it has strong cross-lingual transfer capabilities

High accuracy

Achieves 98.7% accuracy on Catalan POS tagging tasks

Universal Dependencies compatibility

Trained on the Universal Dependencies v2.8 dataset

Model Capabilities

POS tagging

Multilingual text processing

Syntax analysis

Use Cases

Natural Language Processing

Text preprocessing

Provides POS tagging preprocessing for downstream NLP tasks

Improves the accuracy of subsequent tasks such as syntax analysis and named entity recognition

Linguistic research

Used for linguistic research and syntax analysis

Helps researchers analyze the grammatical structure of Catalan

🚀 XLM-RoBERTa base Universal Dependencies v2.8 POS tagging: Catalan

This model addresses the task of part - of - speech tagging in Catalan within the Universal Dependencies v2.8 framework. It is a valuable asset for natural language processing tasks, offering insights into cross - lingual transfer as presented in our related research.

🚀 Quick Start

This model is part of our paper called:

Make the Best of Cross - lingual Transfer: Evidence from POS Tagging with over 100 Languages

Check the Space for more details.

💻 Usage Examples

Basic Usage

from transformers import AutoTokenizer, AutoModelForTokenClassification

tokenizer = AutoTokenizer.from_pretrained("wietsedv/xlm-roberta-base-ft-udpos28-ca")
model = AutoModelForTokenClassification.from_pretrained("wietsedv/xlm-roberta-base-ft-udpos28-ca")

📄 License

The model is released under the Apache 2.0 license.

📊 Model Information

Property	Details
Model Type	Transformer - based model for part - of - speech tagging
Training Data	Universal Dependencies v2.8
Metrics	Accuracy

📈 Model Results

The following table shows the accuracy of the model on various languages:

Language	Test Accuracy
English	86.3
Dutch	87.2
German	79.2
Italian	90.2
French	90.7
Spanish	94.8
Russian	89.1
Swedish	89.5
Norwegian	84.7
Danish	89.3
Low Saxon	53.3
Akkadian	41.0
Armenian	84.7
Welsh	66.0
Old East Slavic	77.4
Albanian	79.2
Slovenian	79.1
Guajajara	32.9
Kurmanji	78.2
Turkish	76.2
Finnish	84.7
Indonesian	84.5
Ukrainian	87.5
Polish	87.4
Portuguese	91.4
Kazakh	80.6
Latin	79.3
Old French	66.5
Buryat	62.8
Kaapor	27.5
Korean	61.6
Estonian	87.2
Croatian	88.8
Gothic	29.1
Swiss German	42.1
Assyrian	17.2
North Sami	41.0
Naija	40.3
Latvian	85.0
Chinese	32.3
Tagalog	72.5
Bambara	29.8
Lithuanian	84.1
Galician	88.8
Vietnamese	65.2
Greek	85.9
Catalan	98.7
Czech	89.3
Erzya	50.9
Bhojpuri	49.7
Thai	43.4
Marathi	82.2
Basque	74.9
Slovak	89.6
Kiche	39.2
Yoruba	28.8
Warlpiri	36.4
Tamil	82.2
Maltese	36.2
Ancient Greek	62.0
Icelandic	83.2
Mbya Guarani	32.6
Urdu	65.2
Romanian	84.8
Persian	76.7
Apurina	37.3
Japanese	19.9
Hungarian	87.2
Hindi	68.8
Classical Chinese	19.2
Komi Permyak	52.6
Faroese	76.4
Sanskrit	38.4
Livvi	64.0
Arabic	79.2
Wolof	38.2
Bulgarian	89.9
Akuntsu	43.4
Makurap	23.3
Kangri	44.9
Breton	63.5
Telugu	85.0
Cantonese	40.5
Old Church Slavonic	57.8
Karelian	73.3
Upper Sorbian	75.8
South Levantine Arabic	64.0
Komi Zyrian	44.2
Irish	67.2
Nayini	50.0
Munduruku	28.8
Manx	35.3
Skolt Sami	41.3
Afrikaans	86.0
Old Turkish	45.7
Tupinamba	36.6
Belarusian	86.0
Serbian	90.4
Moksha	47.7
Western Armenian	78.7
Scottish Gaelic	54.8
Khunsari	47.3
Hebrew	91.7
Uyghur	75.4
Chukchi	34.9

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