Update app.py

app.py

@@ -285,7 +285,7 @@ def respond(
 ):
     llama = Llama("models/madlad400-3b-mt-q8_0.gguf")
     #tokens = llama.tokenize(f"<2ja>{message}")#
-    tokens = llama.tokenize(f"{What is the capital of France?}".encode("utf-8"))
+    tokens = llama.tokenize(f"<2ja>{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):