You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am able to reproduce the tutorial results for MPRA (test R2) only when using GPUs. 2020-05-25 13:18:22,128 - test r2: 0.9223668070860704
When setting use_cuda (forcing CPUs) to False I get a training R2 and near to zero, and negative testing R2 values. 2020-05-22 18:37:47,062 - INFO - test r2: -0.204901966767987
Is there anything that can be checked to allow reproducibility of this example with CPU? Thank you for any comment or suggestion.
Ignacio.
The text was updated successfully, but these errors were encountered:
ilibarra
changed the title
MPRA example - Bad testing R2 with CPU (good with GPUs)
MPRA example - testing r2 with CPU not agreeing with GPUs.
May 25, 2020
ilibarra
changed the title
MPRA example - testing r2 with CPU not agreeing with GPUs.
MPRA tutorial - testing r2 with CPU not agreeing with GPUs.
May 25, 2020
Thanks for posting this issue - it will be good to figure out why CPU doesn't seem to reproduce GPU in your case... what version of Selene are you using? Can you give any more details about your environment/etc that would make it easier for us to reproduce?
Thank you for the development of this package,
I am able to reproduce the tutorial results for MPRA (test R2) only when using GPUs.
2020-05-25 13:18:22,128 - test r2: 0.9223668070860704
When setting
use_cuda
(forcing CPUs) to False I get a training R2 and near to zero, and negative testing R2 values.2020-05-22 18:37:47,062 - INFO - test r2: -0.204901966767987
Is there anything that can be checked to allow reproducibility of this example with CPU? Thank you for any comment or suggestion.
Ignacio.
The text was updated successfully, but these errors were encountered: