automl / DeepCAVE

An interactive framework to visualize and analyze your AutoML process in real-time.
https://automl.github.io/DeepCAVE/main/
Apache License 2.0
69 stars 11 forks source link

Footprint Not Working #191

Open mwever opened 2 weeks ago

mwever commented 2 weeks ago

Description

When trying to generate configuration footprints, some error occurs but it is not visual to me what kind of error...

This is the stacktrace I get when trying to generate the configuration footprint:

deepcave.evaluators.footprint (INFO): Added 20 configurations.
Starting to calculate distances and add border and random configurations...
deepcave.evaluators.footprint (INFO): Starting to calculate distances and add border and random configurations...
Found 100 configurations...
deepcave.evaluators.footprint (INFO): Found 100 configurations...
Added 14 border configs and 135 random configs.
deepcave.evaluators.footprint (INFO): Added 14 border configs and 135 random configs.
Total configurations: 169.
deepcave.evaluators.footprint (INFO): Total configurations: 169.
Getting MDS data...
deepcave.evaluators.footprint (INFO): Getting MDS data...
/Library/Developer/CommandLineTools/Library/Frameworks/Python3.framework/Versions/3.9/lib/python3.9/multiprocessing/resource_tracker.py:216: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown
  warnings.warn('resource_tracker: There appear to be %d '

The error message in the web UI only says: "The job failed. Check the logs or make sure the worker is still running. Most of the times, a simple restart might help."

I am not sure whether this is again a Mac issue or whether this problem also occurs on other platforms. I tried different runs of different folders and the problem is consistently showing up.

mwever commented 1 week ago

I just went back to commit 6169c53befb5f958790b46b84fd3fbffea07d336 from August 2nd 2024. Back then the configuration footprint still worked. So maybe we can isolate the changes turning into the observed bug moving on from there?

mwever commented 1 week ago

Now I found the commit on the main branch that introduced the breaking changes: 7f806aea3d69d4b8f5374f0e0f485b10b45538a7

So for previous commits the configuration footprint still worked. @KrissiHub can you have a look at the diff for these commits to fix the bug, please?