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
start image training, find the x64 process and take down the memory usage.
cancel and switch to another scenario, start training. the memory usage should drop down vastly, but it doesn't
after the new training complete, the memory usage for x64 process is still quite high
We do have to actually CANCEL the training process. It's highly important, especially to compute-intense tasks like image classification. Or it will significantly affect other running tasks. (In my case it almost freeze my laptop even after I hit cancel, and task manager shows both CPU and memory usage are quite high, it doesn't recover until I kill the service process)
The text was updated successfully, but these errors were encountered:
LittleLittleCloud
changed the title
Image training doesn't actually stop after cancel and switch to new scenario
Image training doesn't actually stop after cancel /switch to new scenario
Jan 2, 2020
Thanks for reporting! I agree, lets push on ML.NET to support a true pipeline cancel.
JakeRadMSFT
changed the title
Image training doesn't actually stop after cancel /switch to new scenario
ML.NET needs pipeline cancel (image classification in particular)
Jan 3, 2020
Steps to reproduce:
We do have to actually CANCEL the training process. It's highly important, especially to compute-intense tasks like image classification. Or it will significantly affect other running tasks. (In my case it almost freeze my laptop even after I hit cancel, and task manager shows both CPU and memory usage are quite high, it doesn't recover until I kill the service process)
The text was updated successfully, but these errors were encountered: