Fine-tune with own data #187
Description
Hello, when I used my own data set to fine-tune Swinir on the classic SR task, I found that the L1 loss did not decrease. What is the reason?
{{ message }}
Hello, when I used my own data set to fine-tune Swinir on the classic SR task, I found that the L1 loss did not decrease. What is the reason?