Resuming conversations after reaching rate limit

Whenever you reach the rate limit, why is it that the model is incredibly disoriented afterwards?

When coding with ChatGPT, it almost always starts hallucinating and mixing up old and new code, basically outputting ridiculous irrelevant mixed-up code where functions are and intentions are mixed up.

Is there any way around this currently?

I assume that devs are aware of the issue.