CAMeL-Lab/bert-base-arabic-camelbert-ca-sentiment

1---2language: 3- ar4license: apache-2.05widget:6 - text: "أنا بخير"7---8# CAMeLBERT-CA SA Model9## Model description10**CAMeLBERT-CA SA Model** is a Sentiment Analysis (SA) model that was built by fine-tuning the [CAMeLBERT Classical Arabic (CA)](https://huggingface.co/CAMeL-Lab/bert-base-arabic-camelbert-ca/) model.11For the fine-tuning, we used the [ASTD](https://aclanthology.org/D15-1299.pdf), [ArSAS](http://lrec-conf.org/workshops/lrec2018/W30/pdf/22_W30.pdf), and [SemEval](https://aclanthology.org/S17-2088.pdf) datasets.12Our fine-tuning procedure and the hyperparameters we used can be found in our paper *"[The Interplay of Variant, Size, and Task Type in Arabic Pre-trained Language Models](https://arxiv.org/abs/2103.06678)."13* Our fine-tuning code can be found [here](https://github.com/CAMeL-Lab/CAMeLBERT).14 15## Intended uses16You can use the CAMeLBERT-CA SA model directly as part of our [CAMeL Tools](https://github.com/CAMeL-Lab/camel_tools) SA component (*recommended*) or as part of the transformers pipeline.17#### How to use18To use the model with the [CAMeL Tools](https://github.com/CAMeL-Lab/camel_tools) SA component:19```python20>>> from camel_tools.sentiment import SentimentAnalyzer21>>> sa = SentimentAnalyzer("CAMeL-Lab/bert-base-arabic-camelbert-ca-sentiment")22>>> sentences = ['أنا بخير', 'أنا لست بخير']23>>> sa.predict(sentences)24>>> ['positive', 'negative']25```26You can also use the SA model directly with a transformers pipeline:27```python28>>> from transformers import pipeline29e30>>> sa = pipeline('text-classification', model='CAMeL-Lab/bert-base-arabic-camelbert-ca-sentiment')31>>> sentences = ['أنا بخير', 'أنا لست بخير']32>>> sa(sentences)33[{'label': 'positive', 'score': 0.9616648554801941},34 {'label': 'negative', 'score': 0.9779177904129028}]35```36*Note*: to download our models, you would need `transformers>=3.5.0`.37Otherwise, you could download the models manually.38 39## Citation40```bibtex41@inproceedings{inoue-etal-2021-interplay,42    title = "The Interplay of Variant, Size, and Task Type in {A}rabic Pre-trained Language Models",43    author = "Inoue, Go  and44      Alhafni, Bashar  and45      Baimukan, Nurpeiis  and46      Bouamor, Houda  and47      Habash, Nizar",48    booktitle = "Proceedings of the Sixth Arabic Natural Language Processing Workshop",49    month = apr,50    year = "2021",51    address = "Kyiv, Ukraine (Online)",52    publisher = "Association for Computational Linguistics",53    abstract = "In this paper, we explore the effects of language variants, data sizes, and fine-tuning task types in Arabic pre-trained language models. To do so, we build three pre-trained language models across three variants of Arabic: Modern Standard Arabic (MSA), dialectal Arabic, and classical Arabic, in addition to a fourth language model which is pre-trained on a mix of the three. We also examine the importance of pre-training data size by building additional models that are pre-trained on a scaled-down set of the MSA variant. We compare our different models to each other, as well as to eight publicly available models by fine-tuning them on five NLP tasks spanning 12 datasets. Our results suggest that the variant proximity of pre-training data to fine-tuning data is more important than the pre-training data size. We exploit this insight in defining an optimized system selection model for the studied tasks.",54}55```