NLP Fine-Tuning/Data Labeling - Personal Project
Fine-tuned a pre-trained Hugging Face transformer model for multi-class sentiment classification. Enhanced dataset quality through balancing techniques and optimized learning rate for better outcomes. Developed hands-on expertise in text data preprocessing and model performance evaluation. • Labeled and balanced sentiment classes for classification • Integrated NLP pipeline using Python and Hugging Face • Evaluated model outputs and iterated for improved F1-score • Focused on multi-class text annotation and AI training tasks