数据问题 about unisiam HOT 5 CLOSED

Please make sure that you are able to train, i.e., the running environment is correct, before testing.

from unisiam.

Please make sure that you are able to train, i.e., the running environment is correct, before testing.

您好，以下是我的运行环境，是否可行呢
镜像：Py Torch 1.11.0Python 3.8(ubuntu20.04)Cuda 11.3
GPU：RTX 3090(24GB)*4
CPU：60 vCPU AMD EPYC 7543 32-Core Processor
内存：320GB
在训练和测试都出现了同样的情况，为什么每次运行都是文件或者图像不存在，而且错误情况的图片不固定，可是数据本身没有问题。
测试时，也有将def build_train_loader中的partition='train',改成test。我不太清楚到底哪里出现了问题，不知道您可否帮助我一下，万分感谢。

from unisiam.

Please check your data path (--data_path) carefully. In addition, for testing, you do NOT need to change any code.

Please use English to make your question helpful to others.

from unisiam.

请仔细检查您的数据路径 （--data_path）。此外，对于测试，您无需更改任何代码。

请使用英语使您的问题对其他人有所帮助。

Ok, thank you very much for your answer. I'll try again.
May I ask if I want to change a small data set for debugging (such as WHU-RS19, or is there a suitable small data set recommendation), do I need to change the format and size of the images in the code?

from unisiam.

Please check your data path (--data_path) carefully. In addition, for testing, you do NOT need to change any code.

Please use English to make your question helpful to others.

Thank you. I've found the problem. I'm on the wrong path. On Linux/MacOS, paths use slashes/splits. So -- data_path. /dataset/mini is correct. In Windows, paths are split by backslash , so -- data_path. \dataset\mini should be used.
Now, it's running, and I'm wondering how long does it probably take to run a test? Is it normal to have no reaction for a long time.Thank you for your answer

from unisiam.