This project demonstrates sentiment analysis using ModernBERT, an improved version of BERT designed for better computational efficiency, longer text handling, and enhanced interpretability. The implementation includes fine-tuning ModernBERT on IMDb dataset and evaluating its performance.
🔧 Features ✅ Fine-tuning ModernBERT for sentiment analysis ✅ Training and evaluation using Hugging Face’s Trainer API ✅ Performance evaluation with accuracy & F1-score ✅ Model inference on sample inputs
📦 Dependencies Ensure you have the following dependencies installed:
pip install --upgrade transformers torch datasets scikit-learn
Training process will need wandb api key.
