Zeyu Zhu
Zeyu Zhu
I fix this bug when decreasing the cache_percentage from 0.21 to 0.001. However I am curious on how to realize the cache ratio of 0.21 reported in your paper? Increasing...
Yeah, I use 2 3090 GPUs with 24GB device memory, 1 for sample and 1 for train, the training batch size is 8000, hidden dim is 64 and I generate...
My papers100M dataset downloaded from https://snap.stanford.edu/ogb/data/nodeproppred/.
Thanks! I will try to print the more detailed information.