YeshimebetBayu/Amharic-multi-label-emotion-classification

# Amharic-multi-label-emotion-classification # Multi-Emotion Classification for Amharic Text This repository hosts the official source code, deep learning notebooks, and dataset supporting our manuscript currently in the final review stage. ## 📊 Dataset Overview * **Dataset Name:** Preprocessed Amharic Emotion Dataset * **Size:** 22,000 text instances * **Task:** Multi-label emotion classification using deep learning and fine-tuned Transformer models. ## 📁 Repository Structure * `preprocessed emotion_dataset (2).xlsx` - The complete 22k preprocessed Amharic text dataset. * `multi_label_emotion_classification.ipynb` - Original baseline deep learning model implementation. * `XLM_base.ipynb` - Fine-tuned XLM-RoBERTa Transformer model notebook. * `mBART.ipynb` - Fine-tuned mBART Transformer model notebook. ## 🚀 How to Run the Code 1. Open any of the `.ipynb` notebook files directly in Google Colab or Jupyter Notebook. 2. Ensure you upload the `preprocessed emotion_dataset (2).xlsx` file to your environment runtime path. 3. Run the cells sequentially to reproduce our experimental baseline and Transformer evaluation results.