Replace manual wait with exponential backoff for API rate limiting

[gabriel1628]

Thank you very much @iusztinpaul for this excellent course and high-quality project.
It stands out as one of the few comprehensive end-to-end Machine Learning projects that effectively demonstrates the entire workflow using state-of-the-art techniques and production-grade libraries.

To manage rate limits, you manually increased the wait time between API calls upon hitting the request threshold, like in the __get_quality_score_batch method of the QualityScoreAgent class.

Have you considered implementing exponential backoff for retries, as recommended in the OpenAI API documentation, instead of relying on fixed delays?

[iusztinpaul]

Hello @gabriel1628

Thank you for your support!

Regarding the API calls, you are right, we could have implemented this better, using tools such as tenacity + retries logic. We just wanted to keep it simple and highlight the key ideas. You can always take it from there and add more complexity.

Paul