search

CH-Liang / DSDML

2 stars 0 forks source link

Accuracy is 0.0000 #2

Open Mdahao opened 3 years ago

Mdahao commented 3 years ago

Hello,

thanks for your work. I have a problem about " python train.py --data cub --loss Binomial " . it's Accuracy is 0.0000. and i create train.txt like this: ./CUB_200_2011/images/200.Common_Yellowthroat/Common_Yellowthroat_0055_190967.jpg and so on. I read the code and it's OK. So Is this a problem with the pytorch version， i use 1.9.1 Can the training model be opened？

CH-Liang commented 3 years ago

Hi, You can refer to this link: https://github.com/bnu-wangxun/Deep_Metric/issues/20

By the way, the baseline code is built about 4 years ago. And bnu-wangxun has open source a new coder repo for MS loss and XBM:

repo-1: https://github.com/MalongTech/research-ms-loss

repo-2: https://github.com/MalongTech/research-xbm

You can try them, I think the results maybe better.

Mdahao commented 3 years ago

您好，您可以参考这个链接： bnu-wangxun/Deep_Metric#20

顺便说一下，基线代码是大约 4 年前构建的。bnu-wangxun 为 MS 损失和 XBM 开源了一个新的编码器存储库：

repo-1：https : //github.com/MalongTech/research-ms-loss

repo-2：https : //github.com/MalongTech/research-xbm

你可以试试，我觉得结果可能会更好。

Wow， thank you！