Update GreaterThan_MLP_V1.1_with_FailuresAnalysis.ipynb

GreaterThan_MLP_V1.1_with_FailuresAnalysis.ipynb

@@ -25,6 +25,13 @@ Pairs of decimal numbers are converted into an 8-dimensional array of their digi
 for instance, 10.00 and 09.21 are transformed to [1, 0, 0, 0, 0, 9, 2, 1]. 
 The model's training focuses on predicting whether one number is greater than another through a single binary label.
 
+    test_cases = [
+        ("Simple Greater", 10.00, 9.21), ("Simple Lesser", 5.50, 50.50),
+        ("Decimal Greater", 54.13, 54.12), ("Decimal Lesser", 99.98, 99.99),
+        ("Edge Case: Large Difference", 0.01, 99.99), ("Edge Case: Zero", 0.00, 5.00),
+        ("Tricky: Same Integer Part", 25.80, 25.79), ("Tricky: Crossover", 49.99, 50.00),
+    ]
+
 The MLP baseline model performs remarkably well, achieving over 99.9% accuracy in deciding "GreaterThan". 
 This indicates that the underlying logic of numerical comparison can be learned from raw digits by a simple neural network, 
 provided the input is structured as a fixed-size vector.