EPFL ML Course, group ytf

This organization was created for the second project of the course CS-433: Machine Learning at EPFL. Our goal was to optimize memory usage and inference throughput for the Apertus-8B model by restricting its vocabulary to only handle a few languages. Our models are all pruned versions of the original model, each created following a different approach for selecting the language-specific tokens. The last number in the name of each model is the final vocabulary size.