Chained prompts
A
Abdel Elrafa
It would be great if we can add multiple prompts with the ability to select the model for each prompt. My specific use case would be to clean up what I said using a model that's good at that, and then using another model to translate the text to another language. I find that using a single prompt to do both to not be very effective.
m
m
This would be fantastic.
I would use it by first having a prompt that does all of my transcription cleanup, whether it's fixing stutters, false starts, etc. And then subsequently, I might move into a separate processing mode based on application. Like, for example, I first do a general cleanup on speech. That's prompt one. And then prompt two is "format this for Slack". Or perhaps in another version, it's do that same prompt one, but then put on top of it, format it for email with a new set of instructions for how I deal with email.
S4GU4R0
yes please!!!
Yash Chitneni
+1 this. A lot of documentation across LLMs suggest that leads to the best output.