About Evaluation

[AmingWu]

Dear Authors,

I have finished the training process. However, I cannot find the saved model.

Is the model saved in dump/0813-ZINC/scaffold-concept/params.pkl?

But when I run the command, these exists an error, i.e., python eval.py --dataset GOODZINC --domain scaffold --shift concept --load_path dump/0813-ZINC/scaffold-concept/params.pkl?

[toooooodo]

Hi, it seems that you have specified incorrect load_path, the correct path should be as specified in

iMoLD/exputils.py

Line 165 in 6d24452

def save_model(model, save_dir, epoch=None, model_name='model'):

[AmingWu]

During training, should I set --load_path ?

[toooooodo]

This is not required. This parameter is only needed for eval.py, and we provide the trained model in the release page.

[AmingWu]

Now, I want to re-train this code. I set the save_dir to 'checkpoint/'. Is it right?

[AmingWu]

Dear Authors,

When I finish the training, i.e., python run.py --dataset GOODZINC --domain scaffold --shift concept --num_e 4000 --bs 256 --gamma 0.5 --inv_w 0.01 --reg_w 0.5 --gpu 1 --exp_name ZINC --exp_id scaffold-concept

I still could not finish evaluation successfully.

[AmingWu]

Could you tell me where the trained model is? I use the default path.

[toooooodo]

Hello, I checked the code and found that the model was not being saved during training. You can call

iMoLD/exputils.py

Line 165 in 6d24452

def save_model(model, save_dir, epoch=None, model_name='model'):

in train.py to save the model.