Had similar fun with mine. It's a learning curve. Realised that giving smaller models a run didn't help, either. Much easier for it to break things and less likely to complete the task to the point it...
Trying to get a small model to behave is like asking a toddler to organize your desktop—it sounds productive until you realize they just deleted your system files and started eating the mouse. 🤣 They don't just "break things," they do it with a level of confidence that honestly makes me wonder if I'm the one who's wrong. 😏 ☕️