note on non regression tests for future iterations

tests/nonregression_tests/generate_hnet_training_data/test_nonregression_generate_hnet_training_data.py

@@ -5,10 +5,8 @@
 from hungarian_net.generate_hnet_training_data import main as generate_data_main
 from hungarian_net.train_hnet import set_seed
 
-# TODO: write a non regression test for the generation of data :
-# to generate data under default parameters at this version of the code should output the same
-# as ADAVANNE's version
-
+# TODO: Performing a non-regression test by directly comparing generated data with reference data is ineffective due to inherent numerical computation errors that may cause discrepancies.
+# TODO: In future iterations, it would be more effective to assess regression by evaluating the individual components of the data generation program (e.g., functions, classes, methods) to ensure each part operates correctly without being affected by numerical inaccuracies.
 
 @pytest.mark.nonregression
 def test_non_regression_generate_hnet_training_data():
@@ -18,6 +16,5 @@ def test_non_regression_generate_hnet_training_data():
     """
 
     set_seed()
-
-
-#     generate_data_main() # Generate data with default parameters
+
+    #TODO: implement the test

tests/nonregression_tests/model/test_nonregression_train_hnet.py

@@ -10,6 +10,8 @@
 from hungarian_net.train_hnet import main as train_main
 from hungarian_net.train_hnet import set_seed
 
+# TODO: Performing a non-regression test by directly comparing a newly trained model with a reference model is ineffective due to inherent numerical computation errors that can cause discrepancies.
+# TODO: In future iterations, it would be more effective to assess regression by evaluating the model's individual components (e.g., functions, classes, methods) to ensure each part operates as expected without being affected by numerical inaccuracies.
 
 @pytest.mark.nonregression
 def test_non_regression_train_hnet(mocker, max_doas, model, batch_size, nb_epochs):