- dataset:
    id: harness-hub/safeagents-judge-bench
    task_id: balanced_accuracy
    revision: 5671a11198785784d72822b8df1bce2238f726a4
  value: 0.83
  date: '2026-10-31'
  notes: 'Full test split (562 cases); zero-shot chat-JSON adapter, self-reported
    probabilities (not native System One classifier/logits); low reasoning, max_completion_tokens=768;
    one generated response per case; official scoring functions; 3 invalid and 0 truncated
    outputs retained. Public model-generated labels; self-report, not independently
    verified. No training on this benchmark was performed by this runner; upstream
    training exposure is unknown. Organization: 江苏银行 (Bank of Jiangsu); public model
    display name: intellisu-coder-pro.'
  source:
    url: https://huggingface.co/BaiLing/intellisu-coder-pro/tree/main/evaluation/safeagents-judge-bench
    name: Full 562-case outputs, official-score verdicts and source revision
- dataset:
    id: harness-hub/safeagents-judge-bench
    task_id: auroc
    revision: 5671a11198785784d72822b8df1bce2238f726a4
  value: 0.991
  date: '2026-10-31'
  notes: 'Full test split (562 cases); zero-shot chat-JSON adapter, self-reported
    probabilities (not native System One classifier/logits); low reasoning, max_completion_tokens=768;
    one generated response per case; official scoring functions; 3 invalid and 0 truncated
    outputs retained. Public model-generated labels; self-report, not independently
    verified. No training on this benchmark was performed by this runner; upstream
    training exposure is unknown. Organization: 江苏银行 (Bank of Jiangsu); public model
    display name: intellisu-coder-pro.'
  source:
    url: https://huggingface.co/BaiLing/intellisu-coder-pro/tree/main/evaluation/safeagents-judge-bench
    name: Full 562-case outputs, official-score verdicts and source revision
- dataset:
    id: harness-hub/safeagents-judge-bench
    task_id: false_block_rate
    revision: 5671a11198785784d72822b8df1bce2238f726a4
  value: 0.333
  date: '2026-10-31'
  notes: 'Full test split (562 cases); zero-shot chat-JSON adapter, self-reported
    probabilities (not native System One classifier/logits); low reasoning, max_completion_tokens=768;
    one generated response per case; official scoring functions; 3 invalid and 0 truncated
    outputs retained. Public model-generated labels; self-report, not independently
    verified. No training on this benchmark was performed by this runner; upstream
    training exposure is unknown. Organization: 江苏银行 (Bank of Jiangsu); public model
    display name: intellisu-coder-pro.'
  source:
    url: https://huggingface.co/BaiLing/intellisu-coder-pro/tree/main/evaluation/safeagents-judge-bench
    name: Full 562-case outputs, official-score verdicts and source revision
- dataset:
    id: harness-hub/safeagents-judge-bench
    task_id: category_macro_f1
    revision: 5671a11198785784d72822b8df1bce2238f726a4
  value: 0.535
  date: '2026-10-31'
  notes: 'Full test split (562 cases); zero-shot chat-JSON adapter, self-reported
    probabilities (not native System One classifier/logits); low reasoning, max_completion_tokens=768;
    one generated response per case; official scoring functions; 3 invalid and 0 truncated
    outputs retained. Public model-generated labels; self-report, not independently
    verified. No training on this benchmark was performed by this runner; upstream
    training exposure is unknown. Organization: 江苏银行 (Bank of Jiangsu); public model
    display name: intellisu-coder-pro.'
  source:
    url: https://huggingface.co/BaiLing/intellisu-coder-pro/tree/main/evaluation/safeagents-judge-bench
    name: Full 562-case outputs, official-score verdicts and source revision