TahiraAbbas/bert-multilingual-imdb-sentiment

Multilingual BERT for Sentiment Classification

This model is bert-base-multilingual-cased (an encoder-only Transformer pre-trained on 104 languages), fine-tuned on the IMDB movie-review dataset to classify text as POSITIVE or NEGATIVE. It reads the whole input at once, using the words on both the left and the right, and returns a label with a confidence score.

Although it was fine-tuned on English reviews, it can also classify text in other languages, such as Spanish, French and Urdu.

Details

Item	Value
Architecture	Encoder-only Transformer (BERT)
Base model	bert-base-multilingual-cased
Language	Multilingual (104 languages)
Dataset	IMDB movie reviews
Task	Sentiment classification
Labels	0 = NEGATIVE, 1 = POSITIVE

Usage

from transformers import pipeline

clf = pipeline("sentiment-analysis", model="TahiraAbbas/bert-multilingual-imdb-sentiment")
print(clf("The movie was excellent."))
print(clf("Ce film était terrible."))

Limitations

Trained on a small subset of IMDB reviews. Accuracy on other languages and other domains can be lower than on English movie reviews. Reviews longer than 512 tokens are truncated.