This internship is part of a research effort on continual learning for GenAI methods. The aim is to set up and evaluate experiments that explore how these models can incorporate new knowledge without forgetting prior information, ultimately establishing a new insight into continual learning and setting the baseline for novel methods. You will analyze existing continual learning benchmarks and potentially prepare and curate datasets for a novel benchmark. Additionally, you will implement training loops, data loaders, evaluation metrices and continual learning algorithms in PyTorch. Finally, you will document experimental setups, methodology, and findings to support reproducibility and future research efforts.