Why is the response truncated[:5]?

[wphtrying]

[zhoubiansining]

response = get_proposal(prompt, self.propose_method, self.temperature, self.max_tokens, self.seed,
                        self.max_length,
                        self.truncation, self.do_sample, self.max_new_tokens)

The response is a list of newly generated lines. We truncate this because sometimes the model may generate too many lines for a single step. You can adjust this truncation according to your understanding.