A

Arbertv2

Developed by UBC-NLP
ARBERTv2 is an upgraded BERT model trained on Modern Standard Arabic (MSA) with a corpus of 243GB text, containing 27.8 billion tokens.
Downloads 267
Release Time : 4/11/2023

Model Overview

ARBERTv2 is a deep bidirectional Transformer model for Arabic language, specializing in Modern Standard Arabic processing, particularly suitable for social media text analysis like Twitter.

Model Features

Large-scale Arabic training
Trained on 243GB of Modern Standard Arabic text, containing 27.8 billion tokens
MSA specialization
Specially optimized for understanding Modern Standard Arabic (MSA)
Social media adaptation
Training data includes Twitter text, making it suitable for social media analysis

Model Capabilities

Arabic text understanding
Masked language prediction
Social media text analysis

Use Cases

Natural Language Processing
Arabic cloze test
Predict masked Arabic vocabulary
Example: Can accurately predict 'العربية' in 'اللغة العربية هي لغة العرب'
Social media analysis
Analyze Arabic Twitter content
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase