can you guys add CogVLM2 to the BrillAI platform?
I want to test its pixel coordinate functionality - ie
return the location of the button named "retry" in [[x0,y0,x1,y1]] format
it should return the bounding box in values between 1-1000 in reference to the image,
locating / pinpointing vlm is not something i have seen yet but cogvlm2 seems to be able to do this
Thanks :) and all the best