When using crewai.kickoff, if the output from LLM is too long, the output is cut off midway

[miagi-quyenhn]

I want to check if the LLM API output has been output to the end, get the results to the end, and display the results in the browser.
Returns are missing when I use llm=gpt-3.5-turbo, max_tokens=256
self.llm = ChatOpenAI(model_name="gpt-3.5-turbo", temperature=0.4, max_tokens=256)
Evidence: