Update app.py

app.py

@@ -20,7 +20,8 @@ def load_pipeline():
                     model= model, 
                     tokenizer = tokenizer,
                     max_new_tokens = 50,
-                    top_k = 30, 
+                    top_k = 30,
+                    top_p = 0.7,
                     early_stopping=True,
                     num_beams = 2,
                     temperature = 0.1,
@@ -38,7 +39,8 @@ def chat_interface(inputs):
     
     #result = llm_chain({"input": query, "history": chat_history_tuples})
     result = llm_chain.predict(input = inputs)
-    return result
+    formated_result = result.split('\n')[-1]
+    return formated_result
 
 llm = load_pipeline()
 chat_history = []