Update app.py

app.py

@@ -14,6 +14,7 @@ DEFAULT_MODEL_PATH = model_file
 
 from llama_cpp import Llama
 llm = Llama(model_path=model_file, model_type="mistral")
+llm._token_eos = 7
 
 
 def predict(input, chatbot, max_length, top_p, temperature, history):
@@ -21,7 +22,7 @@ def predict(input, chatbot, max_length, top_p, temperature, history):
     response = ""
     history.append(input)
 
-    for output in llm(input, stream=True, temperature=temperature, top_p=top_p, max_tokens=max_length, ):
+    for output in llm(input, stream=True, temperature=temperature, top_p=top_p, max_tokens=max_length, stop=["<|im_token|>"]):
         piece = output['choices'][0]['text']
         response += piece
         chatbot[-1] = (chatbot[-1][0], response)