A machine learning specialist is handed a dataset of customers’ audio calls with the goal of transcribing those calls. The dataset has pairs of audio recordings and texts where each recording is less than five minutes and a text corresponding to the conversation in that call. A machine learning specialist is given the task of creating a model and training it on this dataset where it could transcribe calls in the future for storing them as text. The specialist decided to use the SageMaker’s Seq2Seq model. Which training instance type would be most suitable for this job?