openvino via optimum-intel is the only way to actually use my Arc GPU right now. But there is always these additional onnx and IR conversion steps. And that also increaes the storage requirements for locally cached models. Also why is the memory requirement so high? a 5GB model, shouldn't take up like 28 GB of ram to convert to IR... also CausalLM which is one of the main tasks ... is not working at all. But I can see that the issue is getting tracked.
@omarnahdi3380Ай бұрын
it says task image-text-to-text is not supported when I try to export the model, is there a solution for it?
@shivaiholi695Ай бұрын
try opening it in the new browser or clear your cache, it will help