ML-Guide: Update information with regard to LUMI

Hi, I found some minor things that might need updating. Thought I collect them here and then I can discuss with @mvsjober and others. I can help with any changes. I wondered if we should mention the https://github.com/Lumi-supercomputer/LUMI-AI-Guide or https://docs.lumi-supercomputer.eu/laif/software/ai-environment/ when it comes to working with the new LUMI AIF containers. It might not be easy to have a nice flow now that (Puhti)/Mahti are different from LUMI. # ml-llm.md ## vLLM multi-node Update when https://github.com/CSCfi/ai-inference-examples/pull/4 goes through https://github.com/CSCfi/csc-user-guide/blob/ef634ae53300bb320499fa58091cf09d04ce12cf/docs/support/tutorials/ml-llm.md?plain=1#L350-L364 # gpu-ml.md ## Update to new containers This is now depreciated https://github.com/CSCfi/csc-user-guide/blob/ef634ae53300bb320499fa58091cf09d04ce12cf/docs/support/tutorials/gpu-ml.md?plain=1#L68-L72 ## gpu energy Is this still fully functional tool? https://github.com/CSCfi/csc-user-guide/blob/ef634ae53300bb320499fa58091cf09d04ce12cf/docs/support/tutorials/gpu-ml.md?plain=1#L375-L380 # ml-multi.md ## Update to new containers This concerns multiple places, but just posting one as example: https://github.com/CSCfi/csc-user-guide/blob/ef634ae53300bb320499fa58091cf09d04ce12cf/docs/support/tutorials/ml-multi.md?plain=1#L373-L375 ## DeepSpeed should be supported on LUMI https://github.com/CSCfi/csc-user-guide/blob/ef634ae53300bb320499fa58091cf09d04ce12cf/docs/support/tutorials/ml-multi.md?plain=1#L927 # ml-workflows.md ## MLFlow in OOD This should be also supported in LUMI https://github.com/CSCfi/csc-user-guide/blob/ef634ae53300bb320499fa58091cf09d04ce12cf/docs/support/tutorials/ml-workflows.md?plain=1#L22-L25 Best, Marlon