pseudo Dice: 0.0

[ltt648]

Hello, I encountered the following problem while using the MSD Test10 colon dataset to segment lesions.
2024-11-27 15:39:30.635207: Epoch 0
2024-11-27 15:39:30.635375: Current learning rate: 0.01
2024-11-27 15:45:23.622632: train_loss 0.0728
2024-11-27 15:45:23.622912: val_loss -0.0007
2024-11-27 15:45:23.622983: Pseudo dice [0.0]
2024-11-27 15:45:23.623044: Epoch time: 352.99 s
2024-11-27 15:45:23.623091: Yayy! New best EMA pseudo Dice: 0.0
pseudo Dice is 0 in first Epoch,although it increase to 70 after 1000 Epochs, it is still very low,how can i do for that?