Hi All,
I’m a student working on a project that uses an Azure OpenAI model to perform certain tasks assigned. My main prompt was stored in vectorDB, and everything was working perfectly when I last tested it about two weeks ago.
I got busy with coursework and didn’t touch the project for a while. When I came back to continue documenting my work, the model suddenly wasn’t behaving the way it used to. It was responding with wrong responses. I checked my code, embeddings, and prompt. There were no errors. I have reinserted my prompt in vectorDB again, but nothing helped. Then the next day, without changing anything, it started working normally again.
Now I’m trying to understand what happened. Does model performance degrade if you don’t use it for a while, or is there another explanation for this kind of temporary behavior change?
Has anyone experienced something similar or knows what might cause this?