Closed Choneke closed 1 year ago
You should use the model weights for initialization. For example,
pretrained_dict = torch.load(model_dict_file)['state_dict']
or use the ema model:
pretrained_dict = torch.load(model_dict_file)['state_dict_ema']
Thank you very much. The problem has been solved by modifying it with the code mentioned above.
Hi, thank you for the great work. I used the pre-trained models to train on my dataset. However, I encountered the following error when running the code.
I use the following code, which also used on other models and it work fine.
Can you help me to solve the issue?