Pytorch-UNet icon indicating copy to clipboard operation
Pytorch-UNet copied to clipboard
verified publisher icon milesial

Time consuming

Open RingZing opened this issue 2 years ago • 5 comments

Although no error is reported, it runs very slowly, and the round progress bar does not advance during training epochs.

RingZing avatar Feb 26 '23 08:02 RingZing

Hi,

  • what GPU do you have?
  • what docker image are you using?
  • what docker flags are you using to run the container?
  • are you sure the training happens on the GPU and not the CPU?

milesial avatar Feb 28 '23 13:02 milesial

I'm having the same issue too.

genggeng-123 avatar May 08 '23 08:05 genggeng-123

Yeah,i also have the same question ,I set the training batchsize to 4, a total of about 4500 pieces of training data, according to the speed of training, training rounds take about four hours or so, this speed is unacceptable, I use a graphics card is 2080Ti, the parameters of the configuration as shown in Figure image

YeYQ22 avatar Aug 25 '23 01:08 YeYQ22

I'm having the same issue too.

feifei-Lee avatar Oct 06 '23 02:10 feifei-Lee