About Pre-trained RVT-B and RVT-B* model On ImageNet-1K

[study0098]

Hello! Thank you for the good work! I'm wondering whether it is possible for you to release the RVT-B and RVT-B* model trained from-scratch on ImageNet-1K instead of pre-training first on ImageNet-21K and then finetune. Thanks for your help!