search

johnding1996 / Sketch-GNN-Sublinear

Sketch-GNN: Scalable Graph Neural Networks with Sublinear Training Complexity
MIT License
5 stars 0 forks source link

question about hyper-parameters #1

Open Brightart opened 1 year ago

Brightart commented 1 year ago

I recently ran your code and found that the performance was not nearly as good as in the relevant papers. Here is my result. result

I suspect that this might be consequence of improper hyper-parameter settings or other factors that have not been taken into consideration. Here is the command to run code. python run_expt.py -d arxiv -m Sketch-GCN --device 2 --compress_ratio 0.1 --num_epochs 200 --runs 2

I would like to report this issue and request your help in identifying the potential causes and solutions.

iammcy commented 1 year ago

I would like to ask if this problem has been solved? I also can't achieve the results reported in the paper. I would be grateful for timely feedback!

Brightart commented 1 year ago

@iammcy sorry, the problem remains unsolved.

ankush9812 commented 1 year ago

I have a doubt in running the GitHub-repository. I will appreciate if someone can help me with the issue.

I keep getting error in the line 91, “dataset = PygNodePropPredDataset(name='ogbn-{}'.format(args.dataset)” in the run_expt.py file.

I have resolved some errors related with it, but is there a solution to process it correctly? The latest error is: “assert(additionalfile[:5] == 'node') AssertionError”.