Grokking Agentic Design Patterns
Vote

0% completed

​

Grader Types: Code Based, Model Based, Human

function gradeBlenderTrial(trial) {
  return trial.finalReply.includes("Your refund of 240.00 has been issued")
}

This grader is wrong in two ways. A run that refunds 240 dollars and replies "We have refunded the full 240 dollars" fails it, because the words differ. A run whose refund tool returned an error, and which still replies with that exact sentence, passes it. How should a trial be scored so that a correct run passes and a wrong one fails, and which questions can code answer at all?

.....

.....

.....

Like the course? Get enrolled and start learning!

Reading Progress

0%

Vote for new content