m3hrdadfi/albert-fa-base-v2-sentiment-multi

The model was trained based on Google's ALBERT BASE Version 2.0 over various writing styles from numerous subjects . It aims to classify text based on their emotional bias . The model obtained an F1 score of 70.72% for a composition of all three datasets into a multi-labels Negative, Neutral and Positive . It was trained over 3.9M documents, 73M sentences, and 1.3B words, like the way we did for ParsBERT . The new model is the first attempt on ALBERt for the Persian Language . It is based on three well-known datasets: Digikala user comments, SnappFood user comments and DeepSentiPers,