Cleanup

The model receives the prompt (and optional system message). The run uses scorer contains_any 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 compute_total_18(lines):
    total = 0
    for j in range(len(lines)):
        line = lines[j]
        total = total + line['qty'] * line['price']
    return total

{
  "expected_contains": [
    "for line in lines",
    "line['qty'] * line['price']",
    "return total"
  ]
}