Fix normal bypass: standardize detection result format

app.py

@@ -266,11 +266,15 @@ class HateSpeechDetectorService:
         # If hate_tokens contain B-HARD, I-HARD, increase label to L2_hate
         if any(bio_label in ["B-HARD", "I-HARD"] for _, _, bio_label in hate_tokens):
             label = "L2_hate"
-        # Construct result message
-        result_msg = f"Detection result: {label}\nConfidence: {sentence_prob:.2f}"
+        
+        # Construct result message with consistent format
+        result_msg = f"🔍 **Detection Result**\n\n"
+        result_msg += f"**Classification:** {label}\n"
+        result_msg += f"**Confidence:** {sentence_prob:.2f}\n"
         if hate_tokens:
-            result_msg += f"\nIdentified hate/aggressive expressions: {hate_tokens}"
-        mitigation = "Performed only detection."
+            result_msg += f"**Identified Expressions:** {hate_tokens}"
+        
+        mitigation = "**Detection Only:** Performed only detection."
         debug_info = {
             'sentence_logits': sentence_logits,
             'sentence_probs': sentence_probs,