Can we use in-context multimodal data for finetuning?

[waltonfuture]

Thanks for your great work! However, it seems that we can only use data that contains one image for SFT. Can we use in-context multimodal data (i.e., containing multiple images) for finetuning?

[qyc-98]

yes, the code supports multi-image finetuning

[waltonfuture]

yes, the code supports multi-image finetuning

Thank you. How should I organize my data for multi-image sft? And how to inference with multi-image?

[haochuan-li]

Same problem here. Any update on multi-image sft?

[waltonfuture]

@qyc-98 Hello! Can you provide some simple examples of in-context inference or SFT? Thanks a lot!

[1SingleFeng]

@qyc-98 I have encountered the same problem. Have you resolved it

[pbarker]

+1 also curious about this