Back to all discussions

Manager   •   over 1 year ago

Use NVIDIA API Endpoints for GPU resources (https://build.nvidia.com/explore/discover)

Hi All

Many people are asking for GPU resources because they don't have them locally. I don't have any to hand out, but there are a few workarounds.

If you are just looking to do inference, then you can use NVIDIA's APIs: https://build.nvidia.com/explore/discover
They give free credits and it's pretty easy to setup.
You can see how to incorporate those APIs into a Workbench Project/Application by referring to the Hybrid RAG Project on GitHub:
https://github.com/NVIDIA/workbench-example-hybrid-rag
You can fork/clone that Project and use it to figure out how to use the endpoints.

If you have a corporate account, you can apply to NVIDIA's LaunchPad: https://www.nvidia.com/en-us/launchpad/data-science/ai-workbench/
Admissions isn't guaranteed, but if you are accepted, it will give you an A40 instance with AIWB already setup.
You get it for 48 hours.
You can see a video on it here: https://www.youtube.com/watch?v=4mfJIkXkQVk&ab_channel=WorldofAI

Sincerely,
Tyler

  • 1 comment

  •   •   over 1 year ago

    I don't have GPU locally so I use workbench with NIM APIs and zero number of local GPU setting. Is it qualified for the challenge?

Comments are closed.