Open moliniao opened 1 year ago
Custom data set (10k), using the author's method, after a round of training, loss_dice and loss_mask have been fluctuating at a certain value, how do you know that it will converge if you continue to train?
Hi, you can draw a loss curve and observe the tendency of loss changing.
Can you provide some code samples?
Custom data set (10k), using the author's method, after a round of training, loss_dice and loss_mask have been fluctuating at a certain value, how do you know that it will converge if you continue to train?