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

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 . We tested three well-known datasets for this task: Digikala user comments, SnappFood user comments and DeepSentiPers in two binary-form and multi-form types . The following table summarizes the F1 score obtained as compared to other models and architectures . Please cite in publications as the following: AlberT-Persian: A Lite BERT for Self-supervised Learning of Language Representations for the Persian Language. [ALBERT Persian] The model is based on a well-,