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
Hi!
I'm analysing a very large dataset, and encountered a problem where during the generation of the posterior sample (post training) I kept running out of GPU memory. I narrowed it down to the fact that all of the results are accumulated in the GPU RAM, and only being transferred to the CPU RAM after analysing all the trials. Solved by changing the line above to: return [(s / num_samples).cpu() for s in sums]
The text was updated successfully, but these errors were encountered:
lfads-torch/lfads_torch/post_run/analysis.py
Line 55 in 6f58d24
Hi!
I'm analysing a very large dataset, and encountered a problem where during the generation of the posterior sample (post training) I kept running out of GPU memory. I narrowed it down to the fact that all of the results are accumulated in the GPU RAM, and only being transferred to the CPU RAM after analysing all the trials. Solved by changing the line above to:
return [(s / num_samples).cpu() for s in sums]
The text was updated successfully, but these errors were encountered: