Does Ollama support vision features?
I'd like to ask if Ollama supports vision features, or what exactly I need to do... I've read the details, but I don't quite understand what to do.
It does but I'm not sure what their process is.. I've never tried it, if you CAN us lmstudio I'd recommend it, but otherwise hopefully someone else can chime in to comment on the support!
Hopefully it's as easy as pulling the HF model and letting it figure everything out?
Yes, Ollama supports vision features.
Text Only:
ollama run gemma3:1b
Text+Vision:
ollama run gemma3:4b
ollama run gemma3:12b
ollama run gemma3:27b
Yes, but the Vision support seems broken on Ollama with your quants @bartowski . I generally prefer your quants so I can use KL variants, but each and every attempt to use them on Ollama results in an immediate EOF error. @ollama could this be related to a differing vision implementation from llama.cpp?