Can you expound a bit about why not? Does it produce bad results? Is it slow to respond? Slow to load? I've been wanting to play around with llamafile-based e

Can you expound a bit about why not? Does it produce bad results? Is it slow to respond? Slow to load? I've been wanting to play around with llamafile-based edge functions but storing even small models in GitHub (for automated deploys) is a terrible and often impossible experience.