Skip to main content

LUMI user coffee breaks

Online

LUMI user coffee break

Join us for a LUMI user coffee break!
Our next user coffee break is on Wednesday, 7th October at 13:00 CEST (14:00 EEST) if you want to discuss any issues you may have with the HPC experts of the LUMI User Support Team, HPE, and AMD via Zoom  https://cscfi.zoom.us/j/65727034273?pwd=VEdtY2trVUVKTEhxajZMbFhETWV2Zz09
Please describe your request with all the details you may find necessary (don’t forget to specify the ticket number if the issue has been reported in the ticketing system) in the dedicated part Questions & Answers of our collaborative document HedgeDoc https://siili.rahtiapp.fi/lumi-coffee-break?both 

Running LLM Inference on LUMI with Aitta

This coffee break, we’ll take a closer look at Aitta, (aitta.csc.fi) for running open-weight (language) models on LUMI without setting up any of the infrastructure yourself. You send a request to an OpenAI-compatible API, and Aitta takes care of getting resources from the scheduler, loading the model, and returning the result.
This talk covers what Aitta is, what it’s suited for (research and prototyping, not production or sensitive data), and how to use it through the API or the web chat interface, with a short demo. We will also show how to get started and where to find the docs and support.

If you need any assistance, please do not hesitate to contact the LUMI User Support Team.

Image: Adobe Stock