Update app.py

app.py

@@ -91,12 +91,15 @@ async def predict(request: Request):
                 top_p=0.9
             )
 
-        response = app.state.models["tokenizer"].decode(
+        full_response = app.state.models["tokenizer"].decode(
             outputs[0], 
             skip_special_tokens=True
         ).strip()
-
-        return {"generated_text": response}
+        
+        # Sadece modelin ürettiği kısmı al (prompt'u çıkar)
+        generated_text = full_response[len(prompt):].strip()
+        
+        return {"generated_text": generated_text}
 
     except Exception as e:
         logger.error(f"Tahmin hatası: {str(e)}")