Unable to complete NVIDIA Clara Train SDK on the AWS Cloud

Healthcare Clara Deploy SDK
#1

Hi,

I am following the steps to install Clara in AWS

When clicking on create the stack, there comes a time in the process that fails and everything is reversed showing the following error message

Error_clara_time
Error_clara_time768×347 33.2 KB

It would be a great help to understand how to resolve these issues.

#3

Hi
Thanks for your interest in Clara Train SDK. Please note we have recently release clara train V4.1 based on MONAI 0.8 which uses PyTorch.
This AWS guide is a bit old referring to Clara V3.1. Can you share your usecase as:

  • Do you want to run AIAA, Train or run inference ?
  • Are you doing this for large team for production or is a simple AWS step up for single user good enough
#4
  1. Train
  2. Since it is a proof of concept , its only for a single user for now
#5

For a simple POC you should simply bring up an AWS with a T4, V100.
You should use clara-train-examples/PyTorch/NoteBooks at master · NVIDIA/clara-train-examples · GitHub
Use the startDocker.sh to only start the clara train container instead of a docker compose with triton.
The getting started notebooks should get you up to speed.

Please note these notebooks are using V4.0. We are currently working on updating them to use the V4.1 sdk released last month