guodongqi / fsl-om Goto Github PK

Hi, I am trying to run your code but cannot get the same result as in your paper. Actually it is far from the performance in your paper. Could you please upload the checkpoint? Thank you so much!

您好：
我在使用mini-imageneat 重現您的performance, 但是training accuracy無法提升是否有其他參數需要設置呢？
我使用以下指令在resnet18上進行測試, “python train.py --n_way 5 --k_shot 5 --k_query 15 --skip False --dataset mini --backbone resnet18 --gpu 0”, training accuracy 維持在0.020 , training loss 維持在4.149, 期待您的回覆 謝謝