Unable to access RAG Workbench Notebook

AI & Data Science NVIDIA AI Workbench
1

Hi Team,

I have installed the Nvidia workbench and am trying to access the clone repository as I want to use Rag application to build notebook code.

It has been building the project for hours and progress shown is still 1/34 and there is no much progress. Please help me on this issue so that I can access the notebook.

PFA

image
image1920×1080 265 KB

Thanks,
Priyanka Hundalekar

2

Hi, thanks for reaching out.

Typically the container pull is dependent on system resources and/or the network your system is on. Are you on a stable and fast network connection on your local machine? If you try to pull the container outside of AI Workbench, is it similarly slow?

3

Hi Edwli,

Yes I am having a stable internet connection, I have tried multiple times and waited for hours but there is no progress on build, it is only showing 1/34 even if i wait for ling time. If its possible for you can we have a quick call to discuss and check on this.

Thanks,
Priyanka Hundalekar

4

I think working locally on Windows needs to go through the WSL virtualization layer which can slow things down. You can verify your container pull speeds by opening the WSL distro and running the docker pull command in there directly. Alternatively you can also work on a remote Ubuntu instance if you have access to one.

$ wsl -d NVIDIA-Workbench
$ docker pull ghcr.io/huggingface/text-generation-inference:2.3.0

You can reach out to schedule a call at workbench-ea@nvidia.com

5

Hi Priyanka

Sometimes the container builds get stuck and there’s a log file that needs to be deleted to move things forward.

Can you try the following?

  1. Go to your .nvwb folder
  2. Go into the project-runtime-info folder
  3. Go into the folder that corresponds to your project. It will have the project name in the folder name.
  4. There will be a file in there with name build-output.building.
  5. Delete that file and try to build the container again.