Examples work, custom prompt for code base generation generate OpenAI Error 400

[MorenoLabs]

Super curious about L2MAC, have been testing the examples from quickstart but other basic requests e.g.

l2mac "create a eda app for csv file" (and other basic ones) throw ultimately after 2-3 steps the following error:

MainProcess| 2024-06-16 10:23:10,726,726 multiprocessing INFO [L2MAC][OpenAI API Request Error] <class 'openai.BadRequestError'> ('Error code: 400 - {'error': {'message': "An assistant message with 'tool_calls' must be followed by tool messages responding to each 'tool_call_id'. The following tool_call_ids did not have response messages: call_5NeB7YpcWw1zVdRgPjozrX0s", 'type': 'invalid_request_error', 'param': 'messages.[5].role', 'code': None}}',) | num_retries: 0 / 10

it keeps retrying but same error.

Any idea how to mitigate that? Researching this got no results.

Thanks for your support!

[samholt]

Thank you for raising this.

Yes, I believe I know how this is arising; it is coming from a missing tool response message. Are you able to share the complete log output?

I'll look into this and get back to you with a fix in the next few weeks at the latest.

[MorenoLabs]

here is the log output file
20240706-193334_log.txt

[ballidev]

have had a similar problem using openAI gpt-4o

tried again with turbo and it has worked

[samholt]

Hi @MorenoLabs @ballidev,

Thank you for your patience. I was able to replicate this bug, and I have now pushed a fix #(2ad39de) that resolves this bug. Thank you for reporting.

I have also improved the overall codebase to make L2MAC more stable,

Feel free to let me know if you have any further issues using it,

Best,
Sam