Characteristics of the training data used in LlaVA

[KansaiTraining]

I would very much appreciate if someone answers me these two questions:

1. What is the training data set used to train the LLaVA model?
2. What is the image size of the images used for training?

I am trying to use LlaVA with some images I have, and they have different dimensions, some very large, some small, so I would like to know what the model was originally trained with

Thanks in advance