Could you release the training data used in the paper?

[Weili-NLP]

trafficstars

Could you provide the training data for finetuning LVLM used in the paper ? or What's the detail procedure to build the training data ?

[zhaowei-wang-nlp]

Thanks for your patience. We uploaded the training file, which organizes the images and prompts in the Llava format, to our huggingface dataset (https://huggingface.co/datasets/ZhaoweiWang/DivScene-DivTraj/blob/main/new_cot_nd_stp8_train_tn5_sr0.25_in4.json)