Fine tune how the LLM is asked about if the email is spam or not. Experiment with what question is asked. How the LLM is prepped for the question. Something other the generic thrown together interaction. Note this issue is only for the initial ask to the LLM, not the reevaluation using the checklist.
Fine tune how the LLM is asked about if the email is spam or not. Experiment with what question is asked. How the LLM is prepped for the question. Something other the generic thrown together interaction. Note this issue is only for the initial ask to the LLM, not the reevaluation using the checklist.