Refactor custom_preprocess() for more robustness

[dkrako]

Info about PR:

Instead of passing the complete tuple with model, x_batch, y_batch, a_batch, s_batch, .. we instead pass a dictionary with the variable names as keys.

If a key matches a variable name like x_batch, it will be overwritten.

If a key refers to a new variable, this variable will be passed instance-wise to evaluate_instance().

If we pass None instead of a dictionary in custom_preprocess, no changes or additional variables will be passed to evaluate_instance.

Minimum acceptance criteria

• All tests pass

[codecov-commenter]

Codecov Report

Merging #167 (05ca1c2) into main (a004ea9) will decrease coverage by 0.11%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #167      +/-   ##
==========================================
- Coverage   94.62%   94.51%   -0.12%     
==========================================
  Files          51       51              
  Lines        2420     2370      -50     
==========================================
- Hits         2290     2240      -50     
  Misses        130      130              

Impacted Files	Coverage Δ
quantus/metrics/axiomatic/completeness.py	94.28% <ø> (ø)
quantus/metrics/axiomatic/input_invariance.py	100.00% <ø> (ø)
quantus/metrics/axiomatic/non_sensitivity.py	100.00% <ø> (ø)
quantus/metrics/complexity/complexity.py	100.00% <ø> (ø)
quantus/metrics/complexity/effective_complexity.py	100.00% <ø> (ø)
quantus/metrics/complexity/sparseness.py	100.00% <ø> (ø)
...s/metrics/faithfulness/faithfulness_correlation.py	95.91% <ø> (-0.17%)	⬇️
...ntus/metrics/faithfulness/faithfulness_estimate.py	98.00% <ø> (-0.08%)	⬇️
quantus/metrics/faithfulness/infidelity.py	96.87% <ø> (-0.10%)	⬇️
quantus/metrics/faithfulness/irof.py	98.07% <ø> (-0.08%)	⬇️
... and 23 more

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

[annahedstroem]

Please see the comments. Some require changes on several places, like update docstrings of Returns and Params etc. Thank you!

[dkrako]

Codecov Report

Merging #167 (9bf05fe) into main (eec1eef) will decrease coverage by 0.11%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #167      +/-   ##
==========================================
- Coverage   94.62%   94.51%   -0.12%     
==========================================
  Files          51       51              
  Lines        2420     2370      -50     
==========================================
- Hits         2290     2240      -50     
  Misses        130      130              

Impacted Files Coverage Δ
quantus/metrics/axiomatic/completeness.py 94.28% <ø> (ø)
quantus/metrics/axiomatic/input_invariance.py 100.00% <ø> (ø)
quantus/metrics/axiomatic/non_sensitivity.py 100.00% <ø> (ø)
quantus/metrics/complexity/complexity.py 100.00% <ø> (ø)
quantus/metrics/complexity/effective_complexity.py 100.00% <ø> (ø)
quantus/metrics/complexity/sparseness.py 100.00% <ø> (ø)
...s/metrics/faithfulness/faithfulness_correlation.py 95.91% <ø> (-0.17%) arrow_down
...ntus/metrics/faithfulness/faithfulness_estimate.py 98.00% <ø> (-0.08%) arrow_down
quantus/metrics/faithfulness/infidelity.py 96.87% <ø> (-0.10%) arrow_down
quantus/metrics/faithfulness/irof.py 98.07% <ø> (-0.08%) arrow_down
... and 23 more

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

Coverage seems to have only decreased because this PR has a lot of deletions, so the absolute number of hits has decreased.
The absolute number of misses has not increased though, just the overall percentage of misses in relation to hits.

[annahedstroem]

Great, let's merge!