Update app.py

app.py

@@ -284,7 +284,7 @@ def respond(
     repeat_penalty: float,
 ):
     llama = Llama("models/madlad400-3b-mt-q8_0.gguf")
-    tokens = llama.tokenize(bf"{message}")
+    tokens = llama.tokenize(f"{message}".encode('utf-8'))
     llama.encode(tokens)
     tokens = [llama.decoder_start_token()]
     for token in llama.generate(tokens, top_k=40, top_p=0.95, temp=1, repeat_penalty=1.0):