Closed OldPanda closed 1 year ago
Thanks for pointing this out! It's a really good suggestion. For torch
, torch-geometric
, torch-sparse
, etc., because of the diversity of environments (OS and CUDA), we would not be able to find the best version for you. Users are required to install by themselves, referring to the install guide. But I will add a requirements.txt
file for the rest of the dependencies.
fixed in 7783b31a69c0f276b5c6f26a7d46763e7e42c03b.
Thanks for addressing it!
Is your feature request related to a problem? Please describe. I tried to run benchmark scripts on my local after installing the repo but failed.
Here's how I setup the environment.
First, I ran
pip install -r requirements.txt
python setup.py install
to install the repo and the dependencies. However, the following errors were complained when I tried to run
python main.py
underpygod/benchmark/
ModuleNotFoundError: No module named 'tqdm'
ModuleNotFoundError: No module named 'torch_geometric’
ModuleNotFoundError: No module named 'pyod’
ImportError: 'NeighborSampler' requires either 'pyg-lib' or 'torch-sparse’
AttributeError: 'DOMINANT' object has no attribute 'decision_scores_'. Did you mean: 'decision_score_'?
where the last one can be fixed by https://github.com/pygod-team/pygod/pull/80, but I still have to install the missing modules with commands
then I can run the benchmark script.
Describe the solution you'd like It would be better to have a dedicated
requirements.txt
file inside folderpygod/benchmark/
containing all the required dependencies.Describe alternatives you've considered N/A
Additional context N/A