Question regarding the base model and Libero dataset coordinates used for fine tuning
#1
by kyungtaepark - opened
Hello,
I am interested in using this model and have a few questions regarding its training details.
I understand that this model is a fine-tuned version of xvla-base with the HuggingfaceVLA/liebero. Could you please confirm if this is correct?
And could you clarify which specific Libero dataset was used for the fine-tuning? Did you use whole dataset? Also, does the data use absolute coordinates for the action space?
Thank you for your help.
All "pc_success=0", please check the model or the implementation.