fix: trust score sorting to higher is better

src/populate.py

@@ -33,9 +33,9 @@ def get_leaderboard_df(eval_results_path, eval_requests_path, cols, benchmark_co
                     print(f"Warning: Column '{col}' missing, adding empty column")
                     all_df[col] = 10.0  # Default to highest risk
                     
-            # Sort by Trust Score (ascending - lower is better)
+            # Sort by Trust Score (ascending - higher is better)
             if auto_eval_column_attrs.overall_risk.name in all_df.columns:
-                all_df = all_df.sort_values(by=[auto_eval_column_attrs.overall_risk.name])
+                all_df = all_df.sort_values(by=[auto_eval_column_attrs.overall_risk.name], ascending=False)
                 
             return all_df