Grokking LLM Evals
0% completed
Chapter Assessment: Code and Reference Graders
This bank checks code and reference graders: choosing the cheapest grader that still catches a failure, code checks for format and rules, checks that run output in a sandbox, comparisons with a reference answer, and key fact checklists. It holds 20 questions, and each attempt draws ten of them. Most questions put you in front of a job ticket assistant output and ask which grader fits, what a verdict means, or what a number comes out to
.....
.....
.....
Like the course? Get enrolled and start learning!
Reading Progress
0%