duweidai
loss value is negative
The ISIC Dice score: 0.8744; The ISIC JC score: 0.7953 [0, 1.4520071135297217, 1.61416205042743, 1.6508494510681295, 1.6697128879217] the best model will be saved at ./saved_models_0920/ISIC2018\folder3\I2U_Net_L/best_score_ISIC2018_checkpoint.pth.tar 0%| | 0/227 [00:00<?, ?it/s]current lr: 4.109423525312737e-05 | Train Epoch: 6 [0/1816 (0%)] Loss: -0.559392
Because they give negative value on the loss function and in some conditions, it take the logarithm.
Start training ... 0%| | 0/227 [00:00<?, ?it/s]current lr: 9.779754323328192e-05 | Train Epoch: 1 [0/1816 (0%)] Loss: 0.301088 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 227/227 [00:35<00:00, 6.33it/s] The average loss:-0.2814 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 260/260 [00:06<00:00, 39.18it/s] The ISIC Dice score: 0.7738; The ISIC JC score: 0.6708 [0, 1.4445990145526038] the best model will be saved at ./saved_models_0920/ISIC2018/folder2/I2U_Net_L/best_score_ISIC2018_checkpoint.pth.tar Training Done! Start testing ******************************************************************** ISIC2018 || start **********************************
=> Loading checkpoint './saved_models_0920/ISIC2018/folder2/I2U_Net_L/best_score_ISIC2018_checkpoint.pth.tar' => Loaded saved the best model at (epoch 1) 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 518/518 [00:15<00:00, 33.45it/s] The mean dice: 0.8497; The dice std: 0.1441 The mean IoU: 0.6685; The IoU std: 0.2437 The mean ACC: 0.9180; The ACC std: 0.1150 The mean sensitive: 0.8541; The sensitive std: 0.2169 The mean specificy: 0.9644; The specificy std: 0.0617 The mean precision: 0.7904; The precision std: 0.2551 The mean f1_score: 0.7698; The f1_score std: 0.2189 The mean Jaccard_M: 0.6685; The Jaccard_M std: 0.2437 The mean Jaccard_N: 0.8874; The Jaccard_N std: 0.1634 The mean Jaccard: 0.6685; The Jaccard std: 0.2437 The mean dc: 0.7698; The dc std: 0.2189 The inference time: 11.3536 ******************************************************************** ISIC2018 || end ********************************** 为什么有一行的进度条不显示
