Answer-first summary for fast verification
Answer: Package your code with Setuptools and use a pre-built container. Train your model with Vertex AI using a custom tier that contains the required GPUs., Set up a Compute Engine VM with all necessary dependencies for training. Then, train your model with Vertex AI using a custom tier that includes the required GPUs.
The most efficient and cost-effective approach is to package your code with Setuptools and utilize a pre-built container, then train your model with Vertex AI using a custom tier that contains the required GPUs. This method is preferred because: - Vertex AI simplifies the training process with its managed services. - Pre-built containers ensure efficient dependency management. - Custom training tiers allow for precise hardware specification. - Setuptools offers ease of use for packaging your code. However, setting up a Compute Engine VM with all necessary dependencies for training and then using Vertex AI with a custom tier that includes the required GPUs is also a viable option, especially if you require more control over the training environment. Therefore, both A and C are correct under different scenarios, but C is generally more efficient for most use cases.
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
No comments yet.
You are tasked with developing an image recognition model using PyTorch based on the ResNet50 architecture for a project that requires processing a dataset of 200k labeled images. The project has strict constraints on budget and requires the training to be completed in the shortest possible time without compromising the model's accuracy. Given these constraints, which of the following approaches would you choose to efficiently and cost-effectively scale your training workload using four V100 GPUs? Choose the best option.
A
Set up a Compute Engine VM with all necessary dependencies for training. Then, train your model with Vertex AI using a custom tier that includes the required GPUs.
B
Create a Google Kubernetes Engine cluster with a node pool equipped with 4 V100 GPUs. Prepare and submit a TFJob operator to this node pool.
C
Package your code with Setuptools and use a pre-built container. Train your model with Vertex AI using a custom tier that contains the required GPUs.
D
Generate a Vertex AI Workbench user-managed notebooks instance with 4 V100 GPUs, and use it to train your model.
E
Both A and C are correct.