About the code of product dataset

[ZulunZhu]

Hi, thanks for your valuable work!

For reason that the arxiv experiment uses the whole batch train loader, it works well. But when I run the code of "GGD_ogbn_product_1epoch", it seems that there is an error in the "blocks = [block.to(free_gpu_id) for block in blocks]". The reason seems that "blocks" contains some list structure that can not be transformed to GPU. I think it might be a potential bug and do you have any advice?

Additionally, I saw the answer to other issues that the code of "Paper100M" has been released, but I didn't find it. Thanks for your patience and I appreciate any feedback.

[zyzisastudyreallyhardguy]

Dear Zhu,

I can successfully run the code from my end. Could you please provide more information about the error message, so that we can solve the problem together? For the ogbn-papers100M, the code is basically the same the the ogbn-products one.

Warmest Regards,
Yizhen Zheng

[ZulunZhu]

I print the error place and it found that the block contains some lists which is not tensors. And then the error is:

Thanks for your feedback.

Zulun