problem of FairDiceLoss
1819707238 opened this issue · comments
1819707238 commented
你们所提出的这个FairDiceLoss似乎忘记了在文中使用? 仅在train_novel.py文件中引入并进行了定义,但似乎并未参与计算?
Yu Tian commented
你们所提出的这个FairDiceLoss似乎忘记了在文中使用? 仅在train_novel.py文件中引入并进行了定义,但似乎并未参与计算?
Please run train_finetune.sh and it will call trainer_novel function and inside the function there is one line to define the FairDiceLoss here: dice_loss = FairDiceLoss(num_classes + 1, n_attr=sens_classes, gamma=args.gamma)