reduced top_p to 0.8

app.py

@@ -45,7 +45,7 @@ def llama_generation(input_text: str,
     output_ids = llama_model.generate(input_ids=input_ids,
                                       max_new_tokens=256,
                                       temperature=0.5,
-                                      top_p=0.9,
+                                      top_p=0.8,
                                       repetition_penalty=2.0)
 
     # Decode