Summary

This workshop focuses on training a language model from scratch using PyTorch, led by Angelos from Eleven Labs' speech-to-text team. The presenter highlights their team's state-of-the-art Scribe V2 transcription model and aims to demonstrate how research engineers build models from the ground up, using minimal libraries and pure PyTorch. Participants will have the option to train a small model locally or via Google Colab, gaining insights into the fundamental processes of developing machine learning models.