enoche / MMRec

A Toolbox for MultiModal Recommendation. Integrating 10+ Models...
GNU General Public License v3.0
367 stars 46 forks source link

problem about freedom.py #22

Closed wenyuzzz closed 9 months ago

wenyuzzz commented 1 year ago

https://github.com/enoche/MMRec/blob/3de1ff73ba1cb32951c1e82077644cdc0436660d/src/models/freedom.py#L81-L83

When we create this graph using GPU, it may cause a problem like this:

sim size: torch.Size([63001, 63001]) sim dtype: torch.float32 knn_ind: tensor([[7585306885142836339, 3991370661542720611, 3486688227326894121, ..., 140594981698912, 140594981698912, 4294967295], [354038449242118, 1271310336001, 8387195064482211950, ..., 7957688336701796217, 8319958742476218724, 7795558767364238639], [ 3304159867, 0, 1, ..., 140588979432688, 140594973553136, 140588979527472], ..., [ 117, 117, 4473802540135219199, ..., 140594981698960, 140594981698960, 4294967295], [ 0, 0, 0, ..., 0, 0, 0], [ 0, 0, 0, ..., 0, 0, 0]], device='cuda:2') knn_ind size: torch.Size([63001, 10])

But if we make this graph through cpu. This bug will disappear.

enoche commented 11 months ago

By default, the graph is created on GPU, no problem on my side, you may check you prepared data: image