LLM-as-judge — when the judge is another model — step 5 of 9
The judge is a function judge_rubric(question, answer, rubric)
that returns a dict like {"passed": True, "critique": "..."}.
Your code needs to read the verdict and branch on the binary
passed field — never on a numeric score.
Fill in the missing condition so the loop prints pass for
passing cases and fail for failing ones.
Expected output:
pass
fail
⌘↵ runs the editor.read, then continue.
LLM-as-judge — when the judge is another model — step 5 of 9
The judge is a function judge_rubric(question, answer, rubric)
that returns a dict like {"passed": True, "critique": "..."}.
Your code needs to read the verdict and branch on the binary
passed field — never on a numeric score.
Fill in the missing condition so the loop prints pass for
passing cases and fail for failing ones.
Expected output:
pass
fail
⌘↵ runs the editor.read, then continue.