Key Implementations:
- Baseline Strategies:
- Naive Bayes Classifier
- Logistic Regression Classifier
- RoBERTa-Based Model:
- Developed a RoBERTa-based model for text classification.
- Top Performance:
- Achieved top 3rd place F1 score on the held-out test dataset (amongst the cohort of 120 teams at the university)
- Hyperparameter Tuning:
- Used the Optuna library for hyperparameter tuning.
The project showcases my ability to build and evaluate NLP models, from simple baselines to more advanced state-of-the-art approaches. I developed this as part of my Master's module - "Natural Language Processing" group project.
If you have any questions or feedback, please don't hesitate to reach out.
A few Important links:
Challenge website - https://competitions.codalab.org/competitions/34344
Link to GitHub repo - https://github.com/Perez-AlmendrosC/dontpatronizeme