A question about the cdvae

[szy120320]

Excuse me, I'm a first-year graduate student. My programming ability is limited. So I have a question.

1. I want to know what the MODEL_PATH is. I tried many paths, but none of them worked. this is the hint:

hydra.errors.MissingConfigException: Primary config directory not found.
Check that the config directory '\Users\12032\Desktop\cdvae-main\cdvae\run.py' exists and readable

[szy120320]

Excuse me, I'm a first-year graduate student. My programming ability is limited. So I have a question.

1. I want to know what the is. I tried many paths, but none of them worked. this is the hint:MODEL_PATH

hydra.errors.MissingConfigException: Primary config directory not found.
Check that the config directory '\Users\12032\Desktop\cdvae-main\cdvae\run.py' exists and readable

I have solved the problem. But I have another problem.

[szy120320]

this is error information: AttributeError: 'CrystGNN_Supervise' object has no attribute 'langevin_dynamics'

I find that the attribute 'langevin_dynamics' exists in the CDVAE class, not CrysGNN_Supervise. So How do I solve this problem?

[szy120320]

@txie-93

[szy120320]

when I execute the recon command, I ran into a problem:

100%|██████████| 2030/2030 [02:43<00:00, 12.44it/s]
Evaluate model on the reconstruction task.
Traceback (most recent call last):
  File "C:/Users/12032/Desktop/cdvae-main/scripts/evaluate.py", line 281, in <module>
    main(args)
  File "C:/Users/12032/Desktop/cdvae-main/scripts/evaluate.py", line 194, in main
    all_frac_coords_stack, all_atom_types_stack, input_data_batch) = reconstructon(
  File "C:/Users/12032/Desktop/cdvae-main/scripts/evaluate.py", line 38, in reconstructon
    _, _, z = model.encode(batch)
  File "E:\anaconda3\envs\cdvae\lib\site-packages\torch\nn\modules\module.py", line 947, in __getattr__
    raise AttributeError("'{}' object has no attribute '{}'".format(
AttributeError: 'CrystGNN_Supervise' object has no attribute 'encode'
batch 0 in 8

[utf]

Hi @szy120320, did you figure out this problem?

[szy120320]

Hi @szy120320, did you figure out this problem?

I tried, but it didn't work

[JianZhou711]

I am probably late and it's already solved, but I worked through it by using the Hydra output from the training instead of prop_models as my model_path which was where I got that error.