Passing instructions only once for a message classifier use case

Yeah if there is no other alternative, have to work on getting some labeled data and using fine-tuned model

1 Like