Support finetuneing using multiple images? 

I want to use the model to process relationship between images in certain domain. But the example in "Visual Instruction Tuning" part in readme is only use one image. I wonder if I can use mutliple images/ image sequence in finetuning? Thanks.