#116 Relative stability

[aaarrti]

[codecov-commenter]

Codecov Report

Merging #136 (5b3d323) into main (80c7e7b) will decrease coverage by 0.61%. The diff coverage is 91.43%.

@@            Coverage Diff             @@
##             main     #136      +/-   ##
==========================================
- Coverage   95.02%   94.40%   -0.62%     
==========================================
  Files          51       54       +3     
  Lines        2411     2683     +272     
==========================================
+ Hits         2291     2533     +242     
- Misses        120      150      +30     

Impacted Files	Coverage Δ
quantus/helpers/utils.py	91.87% <ø> (-0.03%)	:arrow_down:
...tus/metrics/robustness/relative_input_stability.py	88.88% <88.88%> (ø)
...tus/metrics/robustness/relative_ouput_stability.py	89.06% <89.06%> (ø)
...cs/robustness/relative_representation_stability.py	89.70% <89.70%> (ø)
quantus/helpers/tf_model.py	87.75% <93.54%> (-6.37%)	:arrow_down:
quantus/helpers/perturb_func.py	96.03% <100.00%> (+0.03%)	:arrow_up:
quantus/helpers/pytorch_model.py	95.94% <100.00%> (+2.32%)	:arrow_up:
quantus/metrics/robustness/__init__.py	100.00% <100.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[annahedstroem]

OK, we need to discuss the performance gains / pros and cons with jax, separately

[aaarrti]

OK, we need to discuss the performance gains / pros and cons with jax, separately

as discussed on 08.09.22, no jax for now. I've changed the implementation to use numpy only

[annahedstroem]

@aaarrti please resolve these conflicts and then I'll review the PR as requested.

Please also add a description of the PR (just open up a new PR request and you'll get a template automatically populated)

[aaarrti]

It is easier to open a new one