Hello, thank you for showing interest in our work. Regarding the pretrained model weights, they adhere to the license terms of the dataset used to train the respective models.
Thanks for the quick reply. The paper states that for pre-training, the IIT-CDIP dataset (excluding 400k RVL-CDIP ) was used. I believe this dataset is public.
Hi team, thanks for the great work. I just want to know whether the pre-trained models are released in the same license as Apache-2.0 license?