Upload app.py

app.py

@@ -31,7 +31,7 @@ with gr.Blocks() as demo:
         prompt_tokens = tokenizer.encode(msg)
         length_of_prompt = len(prompt_tokens)
         if length_of_prompt > max_prompt_length:
-            msg = tokenizer.decode(prompt_tokens[-max_prompt_length:])
+            msg = tokenizer.decode(prompt_tokens[-max_prompt_length-1:])
 
         history[-1][1] = ""
         for response in client.generate_stream(