mPLUG-DocOwl
mPLUG-DocOwl copied to clipboard
DocOwl2,self.model.init_processor(tokenizer=self.tokenizer, basic_image_size=504, crop_anchors='grid_12'),whats meaning of these params
I run the inference demo on huggingface I wonder whats the meaning of these params ,self.model.init_processor(tokenizer=self.tokenizer, basic_image_size=504, crop_anchors='grid_12') and I try this model on my own data for a easy mission for example extracting words from paper ,and it answered "not answerable" ,how can i fix it,the image resolution is 3672 × 4752 。i try 844 × 1052,not work either