fix: skip system output

app.py

@@ -68,7 +68,7 @@ def invoke(history: HistoryType):
     response = pipe(input_text, do_sample=True, top_p=0.95, max_new_tokens=1024)[0][
         "generated_text"
     ]
-    response = response.split("<|im_start|>assistant\n")[-1].split("<|im_end|>")[0]
+    response = response.split("<|im_start|>assistant\n")[-1].split("<|im_end|>")[1]
     return response