I encountered the following error when running the code with a long prompt: Token indices sequence length is longer than the specified maximum sequence length for this model (263 > 77).