saper0 / revisiting_robustness

The codebase used to generate the reults in the ICLR 2023 paper "Revisiting Robustness in Graph Machine Learning".

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

I have never used seml or MongoDB before. Can someone please tell me hoe to proceed

rohanmahure opened this issue · comments

I installed MongoDB in local system and created conda environment and installed seml in it.
Activate the environment.

I am stuck in running following command.
'''
seml rev_data_collection add config/eval_robustness/csbm/csbm_gcn_lp.yaml start
'''

This is the error I am getting.
OperationFailure: Authentication failed., full error: {'ok': 0.0, 'errmsg': 'Authentication failed.', 'code': 18, 'codeName': 'AuthenticationFailed'}

Hi @rohanmahure, to run the experiments with seml, you need to configure seml to use your local MongoDB installation by running seml configure. (See also the seml-documentation here.)

Furthermore, as you run the experiments locally and not on a cluster with a slurm workload manager, you need to append the --local flag, i.e. run

seml rev_data_collection add config/eval_robustness/csbm/csbm_gcn_lp.yaml start --local

instead.

However, I want to mention that you can do experiments without using a MongoDB or seml by directly calling the run method of the exp_eval_robustness.py file as illustrated in the following jupyter notebook. Here, the configuration is just directly passed as dictionaries to the run method.

Let me know if this could solve your issue!