Morgan MG Model Review Chair 1 Scope trust 2 Show calibration 3 Name stop rule Review board Morgan challenges the use of an LLM grader for reinforcement fine-tuning. If the grader is weak, the tuned model may optimize a shortcut. 120 calibration cases 6 hack probes Grader discipline Smooth score, balanced cases, hack probes
Read the full lesson
Sign up free: one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in