Files

pert.py

@@ -44,7 +44,8 @@ perturbation_descriptions = {
     "Exact Name Class Variable Modification": "Renames classes and variables to a specific randomly generated name.",
     "Casing Class Variable Modification": "Generates lexical variations of class and variable names with different casing.",
     "Onecase Modification": "Generates lexical variations of class and variable names with just 1 letter uppercase wither for class anme or variable name.",
-    "Example Modification": "Generates lexical variations of class and variable names with Example being the class name and example being the variable name or vice versa."
+    "Example Modification": "Generates lexical variations of class and variable names with Example being the class name and example being the variable name or vice versa.",
+    "Finetuned on compile error": "Clusters based on finetuned codebert model on compile errors"
 }
 
 def compute_and_log_csi(file_orig, file_pert, perturbation_name, output_csv="results/csi_summary.csv"):
@@ -162,4 +163,11 @@ compute_and_log_csi(
     perturbation_name="Example Modification",
     output_csv="results/csi_summary.csv"
 )
+
+compute_and_log_csi(
+    "codenet_4000_finetuned_compile_error/java/layer12/kmeans/clusters-kmeans-350.txt",
+    "codenet_4000_finetuned_compile_error/java/layer12/kmeans/clusters-kmeans-350.txt",
+    perturbation_name="Finetuned on compile error",
+    output_csv="results/csi_summary.csv"
+)
 # You can now call compute_and_log_csi again and again for other perturbations!

results/csi_summary.csv

@@ -11,3 +11,4 @@ Exact Name Class Variable Modification,0.675121649837896,0.324878350162104,Renam
 Casing Class Variable Modification,0.6722713965133429,0.3277286034866571,Generates lexical variations of class and variable names with different casing.
 Onecase Modification,0.665697304921991,0.334302695078009,Generates lexical variations of class and variable names with just 1 letter uppercase wither for class anme or variable name.
 Example Modification,1.0,0.0,Generates lexical variations of class and variable names with Example being the class name and example being the variable name or vice versa.
+Finetuned on compile error,1.0,0.0,Clusters based on finetuned codebert model on compile errors