Why did it appear killed when training the model

JinheonBaek / RGCN

Pytorch implementation of a RGCN Link Prediction Model

229 stars 41 forks source link

Closed ReadyTeresa closed 3 years ago

JinheonBaek commented 3 years ago

Thank for your interest.

Can you specify the detail of errors for training? Then, I will check it.

Sincerely, Jinheon Baek

ReadyTeresa commented 3 years ago

Thank for your interest.

Can you specify the detail of errors for training? Then, I will check it.

Sincerely, Jinheon Baek

No errors are printed out, but the server will be killed when executing the valid function at line 80

JinheonBaek commented 3 years ago

I will first check the error, and then let you know. Thank you for sharing the issue.

sunke-github commented 3 years ago

It caused by out of memory. When the model begins evaluation, It needs about 64g memory.

JinheonBaek commented 3 years ago

It seems, as @sunke-github pointed out, the error is caused by out-of-memory, not the memory for GPU but the memory for CPU.