Load dataset takes fairly long

[SelenaClover22]

I cannot access the stanford website for the ModelNet40 dataset, so i found one on kaggle and use it instead. However, the train process would stick on load dataset phase for over 30 mins, and then shows some error. I check the pretrained log files and saw that the load time is very small, within one minute.

[13717630148]

have you solved the problem?

[Lobbyism]

Are you sure it's a problem with the dataset? Initially, I tried to use the use_cpu flag. Upon seeing that the training worked, I

1. Created a new environment with a specific Python version
   conda create --name pointnet_env python=3.10
2. Installed dependencies
   conda install pytorch==2.1.2 torchvision==0.16.2 torchaudio==2.1.2 pytorch-cuda=11.8 -c pytorch -c nvidia
3. Downgraded numpy
conda install numpy=1.26

train_classification.py now appears to work. For information, I am using a 3060 Ti.