Hello, your code repository has been very helpful for me to understand the training and fine-tuning process of your work. But I still have some questions. I noticed that the val datasets you used for fine-tuning are S9 and S10 of DIP, while the test datasets used for testing are also S9 and S10. These are the conclusions I got from your preprocess.py, but will such a fine-tuning process cause the model to want to over-understand the data distribution of S9 and S10 during training?
Is there any previous work that did this? Looking forward to your reply!
Hello, your code repository has been very helpful for me to understand the training and fine-tuning process of your work. But I still have some questions. I noticed that the
val datasetsyou used for fine-tuning are S9 and S10 of DIP, while thetest datasetsused for testing are also S9 and S10. These are the conclusions I got from your preprocess.py, but will such a fine-tuning process cause the model to want to over-understand the data distribution of S9 and S10 during training?Is there any previous work that did this? Looking forward to your reply!