{
  "n_tasks": 100,
  "k": 8,
  "pass@1": 0.771,
  "oracle": 0.89,
  "selectors": {
    "pass@1 (random single)": {
      "accuracy": 0.771,
      "est_tokens_per_task": 120
    },
    "visible-only (first visible-pass)": {
      "accuracy": 0.85,
      "est_tokens_per_task": 960
    },
    "no-think verifier (max P_A)": {
      "accuracy": 0.8,
      "est_tokens_per_task": 960
    },
    "thinking verifier (max P_A)": {
      "accuracy": 0.86,
      "est_tokens_per_task": 4960
    },
    "visible + no-think verifier": {
      "accuracy": 0.87,
      "est_tokens_per_task": 960
    },
    "visible + thinking verifier": {
      "accuracy": 0.87,
      "est_tokens_per_task": 4960
    },
    "ORACLE pass@k (non-deployable)": {
      "accuracy": 0.89,
      "est_tokens_per_task": 960
    }
  },
  "c2_false_pass": {
    "visible_pass_rate": 0.818,
    "false_pass_rate_among_visible": 0.066,
    "n_false_passes": 43,
    "among_visible_passers_verifier_auroc_think": 0.758,
    "among_visible_passers_verifier_auroc_nothink": 0.701
  }
}