Hey everyone, I’m using the GPT-4o model with the Assistant API, and I’ve been running into a frustrating repetition issue. Sometimes GPT-4o repeats the same function calls or prompt until it hits the max_tokens limit. I tried tweaking temperature (it was 0.2 before and is now 0.7), but the repetitive output still happens.
Has anyone else experienced this repetition problem like this? If so, how did you resolve it (or at least reduce it)? Any insights on potential configuration or prompt engineering tricks would be really helpful! Thank you!