Raised a bug about #ramalama today not playing well with #arm64 and and #amd gpus. However if you force the base image local inference does use
@stsquad@mastodon.org.uk @stsquad@mastodon.org.uk @stsquad@mastodon.org.uk