As the leader of a data science team in a global corporation, you're tasked with optimizing the team's Google Cloud compute costs without compromising the performance of your machine learning models. Currently, the team utilizes high-level TensorFlow APIs on AI Platform with GPUs for model refinement, a process that can take weeks or months. Given the constraints of budget, time, and the need for model effectiveness, which of the following strategies would you recommend? (Choose two correct options)