Chengming Xu

> I guess that something was changed in 'congfig.py' accidentally. > Try the old version. > `git checkout a3e79f8` Yes the branch you mentioned doesnt have this problem. Now i...

You can simply create a 'logs' directory. I think this will work.

> when I run train.py i got this error: > > ModuleNotFoundError: No module named 'torch.nn._functions' > > and i didn't find this module in pytorch's ducument Hi, can you...

I have checked that pytorch 0.4.0 has this module but the latest version has changed that. I think changing your version to 0.4.0/updating the code according to your version (e.g....

> I figure out that you are using the data format from DeepEMD where all images are 224*224. Yes, we use the data provided by DeepEMD with 224x224 size, but...

Hi, unfortunately I cannot remember the exact environment I used when conducting this research. But you can first create an environment that is compatible with Pytorch0.4. I think such an...

Also, when I use the default script for episodic training the usage of RAM increases dramatically during training. The model can use about 100G RAM after about 300 iterations. I...

Hi, thank you for your reply! I have tried to modify the training script based on your new version to skip the model forward and backward and only iterate the...

I re-clone the repo and trained the protonet with your original code. After 1400 iterations 23G RAM is used. When I train the model with 4 GPUs (by modifying the...