Training time on PCQM4M-LSC

snap-stanford / ogb

Benchmark datasets, data loaders, and evaluators for graph machine learning

https://ogb.stanford.edu

MIT License

1.93k stars 397 forks source link

Closed mengliu1998 closed 3 years ago

mengliu1998 commented 3 years ago

Hi,

Thanks for presenting this great dataset and challenge.

Could you provide the rough training time of baselines (e.g., GCN and GIN) on PCQM4M-LSC dataset?

Thank you in advance.

weihua916 commented 3 years ago

Hi, it takes 1 to 1.5 days on a single GPU.

mengliu1998 commented 3 years ago

Hi,

Thank you for the immediate response.

Do you mean the time of training 100 epochs on whole training set? (I guess this is similar to training 1000 epochs on 10% training set.)

weihua916 commented 3 years ago

It is time for training 100 epochs.

mengliu1998 commented 3 years ago

Got it. Thank you for the clarification.