-
Hi there! I’m building applications that use OpenAI’s support for function calling. In the same way as you might test multiple variants of a prompt to find the prompt that works best, I’m looking for a way to test multiple variants of a function, since functions also need to be “prompt engineered” and o work well. Is that something that Prompt Flow can support? I didn’t find anything relevant in the docs. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 6 replies
-
Hi @timrogers , it's supported to tune function calling with variants. Just like other variants, you can create different variants for same LLM node with different function inputs like this: |
Beta Was this translation helpful? Give feedback.
Hi @timrogers , it's supported to tune function calling with variants. Just like other variants, you can create different variants for same LLM node with different function inputs like this: