Computer shuts down when training a model

[SATURN10Participant]

Hi there,

I have experienced a few computer shuts down while training my SAEHD model with DFL.

The training could go on for a few hours then shuts down suddenly. I have tried to lower down the batch size from 36 to 24 as I thought it might relate to OOM issue, however problem still exists.

My computer build:

7800x3d + rtx 4090 + 1050 psu

I have undervolt my gpu to .900v at 2600mhz.

Note: my cpu does get pretty hot when training. (85c)

Thanks a lot

[ChechgroupParticipant]

you need to increase the page file

[Author]

Posts