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
Not a bug, but a consequence of the choices we've made a long time ago: when training with multiple files, CAREamics will instantiate PathIterable datasets for both training and validation.
PathIterable randomly fetch patches from the images, yielding different validation patches each epoch, which in itself is not how we would want to assess training progress.
Describe the bug
Not a bug, but a consequence of the choices we've made a long time ago: when training with multiple files, CAREamics will instantiate
PathIterable
datasets for both training and validation.PathIterable
randomly fetch patches from the images, yielding different validation patches each epoch, which in itself is not how we would want to assess training progress.@CatEek do you agree with the diagnostic?
The text was updated successfully, but these errors were encountered: