Loop Cleanup

The model receives the prompt (and optional system message). The run uses scorer contains_all with the JSON configuration below. Pass/fail and partial credit are determined entirely by that scorer against the model output; no human grading.

Refactor without changing behavior. Return only Python code.

def names_upper(users):
    out = []
    for i in range(len(users)):
        out.append(users[i]["name"].upper())
    return out

{
  "expected_contains": [
    "for user in users",
    "user[\"name\"].upper()",
    "return out"
  ]
}