This model is designed for emotion recognition in Persian text. It predicts the presence of six emotions: anger, disgust, fear, sadness, happiness, and surprise, as well as the primary emotion within these categories, including an "other" category for cases when none of the specified emotions are present. The model leverages XLM-RoBERTa, a pre-trained transformer-based language model, fine-tuned on two datasets: EmoPars and ArmanEmo. It includes a Bidirectional Gated Recurrent Unit (BiGRU) layer to better capture contextual dependencies, improving performance on emotion classification tasks.

This model can be directly used for emotion recognition in Persian text. It can predict the presence of six emotions and the primary emotion from those six or an "other" category. The model checkpoints can be downloaded for local usage or integrated into existing systems.

The model can be further fine-tuned for specific tasks or integrated into larger applications such as sentiment analysis systems, chatbots, and customer service systems where emotion recognition is required.

This model should not be used for tasks involving languages other than Persian or general-purpose sentiment analysis without further adaptation.

Users should be aware of the model's limitations and biases, especially in high-stakes applications such as mental health or sensitive customer feedback. For critical applications, it is recommended to combine this model with other validation tools.