Developer Guide¶
For Machine Learning development¶
Auptimizer makes it a little easier to debug your experiment / job in the following ways.
Set level of logging¶
Logging can be activated using the --log
flag. (E.g. python -m aup <experiment> --log <level>
).
The following logging levels are available:
error - everything stops the process
warn - using default values
info - progress update
debug - everything else
Test in passive mode¶
Change resource
in experiment.json
to "passive"
and then run:
python -m aup <experiment config>
By doing so, Auptimizer will run in a passive mode where it interactively prints running script with its working path and asks for the returned value. You should run your script in a second terminal to see whether it finishes correctly. And then you can return that value back to Auptimizer's command line.
For Auptimizer Software Development¶
Environment¶
Either use virtualenv
:
virtualenv testenv
source testenv/bin/activate
pip install -r requirements.txt
export PYTHONPATH=`pwd`:$PYTHONPATH
or:
export PYTHONPATH=<repo folder>:$PYTHONPATH
Unit testing¶
If you make changes to the Auptimizer code, you can run the included unit tests to make sure that you didn’t break anything.
If it’s the first time you are running these tests, do:
chmod u+x tests/EE/test_Job.py
to set the correct permissions. You can then run the tests using:
python -m unittest