QuestionQ60

VerifiedScaling prototypes into ML models

You are training an object-detection machine learning model using a dataset of three million X-ray images, each approximately 2 GB in size. You use Vertex AI Training to run a custom training application on a Compute Engine instance with 32 cores, 128 GB of RAM, and one NVIDIA P100 GPU. You observe that model training is taking a very long time. You want to reduce the training time without compromising model performance. What should you do?

Explanation

Distributed training divides data-parallel training work across multiple workers and GPUs, reducing wall-clock training time while preserving the training objective. Vertex AI supports distributed TensorFlow training with the tf.distribute.Strategy API, including multi-worker strategies that coordinate data distribution and gradient updates across machines.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!