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
Originally posted by davideNasi August 29, 2022
I'm using torch implementation of TabNet (https://github.com/dreamquark-ai/tabnet) and when I run simulations each client write a 1.6 GB file every epoch in C:\Users<user>\AppData\Local\Temp\ray\session_\ray_spilled_objects.
Am I doing something wrong or is it possible to prevent this?
Current workaround
Downgrade ray to version 1.11.1 : pip install ray==1.11.1
Possible solutions
Find a way to disable object spilling in the current version of ray
Discussed in https://github.com/adap/flower/discussions/1394
Originally posted by davideNasi August 29, 2022
I'm using torch implementation of TabNet (https://github.com/dreamquark-ai/tabnet) and when I run simulations each client write a 1.6 GB file every epoch in C:\Users<user>\AppData\Local\Temp\ray\session_\ray_spilled_objects.
Am I doing something wrong or is it possible to prevent this?
Current workaround
Downgrade
ray
to version1.11.1
:pip install ray==1.11.1
Possible solutions
ray
ray
version to1.11.1
in thepyproject.toml
The text was updated successfully, but these errors were encountered: