customized reward model

#3
by jiang0203 - opened

Thank you for your contribution.
I'm wondering if I can use my pretrained Deeplabv3 or Upernet models in your training script.
Because my dataset is not suitable for the standard pretrained models that are trained on landscape or portrait datasets, I need to train my own reward model.

Thank you for your contribution.
I'm wondering if I can use my pretrained Deeplabv3 or Upernet models in your training script.
Because my dataset is not suitable for the standard pretrained models that are trained on landscape or portrait datasets, I need to train my own reward model.

Sure, as long as your model weights is applicable with mmdet/seg format. You can refer to the github repo for training code.

Sign up or log in to comment