fix: pick best device (CUDA → MPS → CPU) for LoRA model loading#1303

planetf1 merged 2 commits into

generative-computing:maingenerative-computing/mellea:mainfrom

planetf1:issue-1295planetf1/mellea:issue-1295Copy head branch name to clipboard