Closed amirshnll closed 1 year ago
Hi amirshnll,
This import code serves for a unique feature: You can create any new dataset classes and register them without any modifications of the __init__.py
file. Using this repo won't cause this issue since the code always passes the training test.
Can you tell me which GOOD version you are using and post your error log for me to check? I would like to help you with this issue. 😄
Best, Shurui
I used the repository sample code in documentation:
https://good.readthedocs.io/en/latest/tutorial.html#good-datasets
but not in a personal dataset.
I have just tried the code:
# Directly import
from GOOD.data.good_datasets.good_hiv import GOODHIV
hiv_datasets, hiv_meta_info = GOODHIV.load(dataset_root, domain='scaffold', shift='covariate', generate=False)
# Or use register
from GOOD import register as good_reg
hiv_datasets, hiv_meta_info = good_reg.datasets['GOODHIV'].load(dataset_root, domain='scaffold', shift='covariate', generate=False)
cmnist_datasets, cmnist_meta_info = good_reg.datasets['GOODCMNIST'].load(dataset_root, domain='color', shift='concept', generate=False)
and didn't meet the problem you mentioned.
Please let me know if the issue persists.
Hello @divelab, n my opinion, this repository's code is amazing and very clean. It is very helpful to use others. My case involves a
Circular Import error
bug, which I fixed by changing two files in this repository:GOOD/data/__init__.py
file i changedto
GOOD/data/good_datasets/__init__.py
file i changedto
Please update the repository if you believe this is a global issue.