Evaluation and fine-tuning a low-performing model

In this repo I try to put into practice what I learned from the "Evaluation and light customization of large language models" course from Nvidia following the python best practices.