fix(pd): keep a vision-capable decode leg for multimodal requests - #2655

fix(pd): keep a vision-capable decode leg for multimodal requests#2655

hello-alexmcc wants to merge 1 commit into

mainsmg-project/smg:mainfrom

fix/pd-vision-capable-decode-selectionsmg-project/smg:fix/pd-vision-capable-decode-selectionCopy head branch name to clipboard