Llava style model support in Onnx?

[kinchahoy]

Hey folks,

I was thinking about training a small Phi-3 + OpenClip Llava style model (building off the MobileVLM work). Do you plan to support that sort of multi-modal model with Onnx (or ideally onnxruntime-genai)? Let me know if I can help test etc.

[natke]

@kinchahoy Yes we do have this support on our road map. I am moving to our Discussions forum to track it there