IGB-Datasets

IGB-Datasets copied to clipboard

fix: data already downloaded

1

jasperzhong

How to reproduce accuracy results for IGB-large

4

I am unable to reproduce the accuracy results in the paper for IGB-large+SAGE model. I got 60~61% validation and test accuracy after 3 epochs, compared to 64.89% in the paper....

hirayaku

I think this might be a bug. Could someone take a look if it's convenient? When I try to modify https://github.com/IllinoisGraphBenchmark/IGB-Datasets/blob/main/igb/train_single_gpu.py#L150 on the small and medium datasets, there's always an...

yichuan-w

The training set accuracy is much lower than the test set accuracy.

31

Provide context and explain with the example if possible and relevant. BTW, Why is the training node accuracy for IGB much lower than the validation and test accuracy? Usually, based...

yichuan-w

Currently, IGB dataloaders use numpy mmap. numpy mmap is extremely slow and single-threaded. We should consider enabling torch mmap to enable multithreaded support when reading the data from the disk....

msharmavikram

We should consider adding the dataset to the huggingface dataset repository for wider usage. Details on how to add the dataset is available here - https://huggingface.co/docs/hub/datasets-adding

msharmavikram

akhatua2

Download of large dataset is quite slow

1

Hi! I'm trying to download the `IGB-large` dataset. but I found that it's very slow to download the raw feature file (say .npy file). Do you have any plan about...

zhuangbility111