Reproduction Issue

[YUjh0729]

Hi! Thanks for sharing your work.
I replicated the paper's code, but the results astonished me. My training and testing Dice results exceeded those in the original paper by a significant margin, and I don't know why! I used the 2D Abdomen MR dataset and training environment provided by the paper. I suspect that the training process automatically loads the authors' pre-trained model, allowing the results to achieve great performance after about 30 epochs.
Could you provide the progress.png image of your training results?

[moeinheidari7829]

Hi! Thanks for sharing your work. I replicated the paper's code, but the results astonished me. My training and testing Dice results exceeded those in the original paper by a significant margin, and I don't know why! I used the 2D Abdomen MR dataset and training environment provided by the paper. I suspect that the training process automatically loads the authors' pre-trained model, allowing the results to achieve great performance after about 30 epochs. Could you provide the progress.png image of your training results?

Hi, this is the same thing I am experiencing too. Did you found what the reason is?