brittle hemlock
Hey,
I tried changing my existing app to the new model (gpt-4-1106-preview) and something seems deeply broken with function calling / JSON responses - see attached images.
For clarification, the only thing I have done is changing the model from gpt-4-0613 to gpt-4- 1106-preview in my code, nothing else
fierce jetty
Yes, I have similar issues, with every single prompt on the new preview model. It's just seems to ignore random parts of prompts. No matter if I write it as a rule into the prompt, as IMPORTANT, or whatever. It just seems to love regenerate the whole answer over and over again instead of following given guidelines.
brittle hemlock
@fierce jetty are you also seeing issues with function calling?