Why some json_mode is set to False when calling LLM?

[Hanson13]

❓ Questions and Help

We sincerely suggest you to carefully read the documentation. After that, if you still feel puzzled, please describe the question clearly under this issue.

Hi, I thought almost all LLM responses should be in json format and setting json_mode to True can bring better reliability, but I am confused that some json_mode is default or set to False when calling LLM. Could you kindly tell me the reason or when should I set json_mode as False?

Thank you very much for your response!

[Hoder-zyf]

I think if you want the response to be in JSON format, you can always set json_mode to True. As you can see in rdagent/oai/backend/base.py (lines 548-549), json_mode is actually a custom parameter we defined ourselves:

if response_format is None and json_mode:
  response_format = {"type": "json_object"}

If you want a more comprehensive understanding, you can refer to the different approaches shown in test/oai/test_completion.py—as we support all of them.

[Hanson13]

I think if you want the response to be in JSON format, you can always set json_mode to True. As you can see in rdagent/oai/backend/base.py (lines 548-549), json_mode is actually a custom parameter we defined ourselves:

if response_format is None and json_mode:
response_format = {"type": "json_object"}
If you want a more comprehensive understanding, you can refer to the different approaches shown in test/oai/test_completion.py—as we support all of them.

Thanks for your answer!