{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/da4ad333a5744f02852407997dfda181\" frameborder=\"0\" width=\"1130\" height=\"847\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":847,"width":1130,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":847,"thumbnail_width":1130,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/da4ad333a5744f02852407997dfda181-1689274669369.gif","duration":180.25,"title":"Terracotta: Easily Train and Experiment with Large Language Models 🚀","description":"Hey there! In this video, I'll walk you through how to use Terracotta, a platform designed to make training and experimenting with large language models a breeze. First, I'll show you how to upload your data, which is the heart of any machine learning model. It's as simple as selecting a CSV file and deciding whether you want to use your own validation data or have it automatically generated. Once your data set is uploaded, you can explore all the fields, examine the train and validation data, and fine-tune your model on that specific data set. You have the freedom to choose the model architecture, format the input to your liking, and specify hyperparameters. And the best part? Launching a fine-tuning job is just a click away! After training your model, you'll be able to analyze a training loss curve, delve into meta information about the model, and review valuable metrics. So, let's dive in and make the most out of Terracotta together! 🌟"}