is right the code of multi-image

[i-zhangjingjun]

thanks your project, i learned the code ,but i can't find difference between the origin liuhaotian’s LLaVA and your file of LLAVA, can you introduce the difference and which code can train multi-image ? thank you very much

[Liuziyu77]

https://github.com/Liuziyu77/MMDU/blob/main/LLaVA/llava/train/train.py#L724
You can refer to here.