Sagemaker deployment failure

#4
by xiaoweiwen - opened

Hello everyone,

I was trying to deploy the model as sagemaker endpoint, with ml.g5.12xlarge instance (which has 96gb on GPU) but it fails with cuda out of memory error.
Follows the screenshot of the Cloudwatch error log.

image.png

Same error here with same EC2 instance

Sign up or log in to comment