continuation_benchmark/both_metric_eval/summary.json

{
  "version": "both_metric_eval_score_v1",
  "generated_utc": "2026-05-03T11:01:11.063073Z",
  "n_plasmids": 47,
  "n_comp_windows": 702,
  "n_surp_windows": 208,
  "per_archetype_completion": [
    {
      "archetype": "BD_patent_sparse",
      "n": 12,
      "Base_mean": -7.9388,
      "SFT_mean": -6.3137,
      "RL_mean": -8.4935,
      "RL-SFT": -2.1798,
      "RL-Base": -0.5547,
      "median_RL-SFT": -1.8541,
      "RL_gt_SFT": 1,
      "RL_gt_Base": 3,
      "pct_RL_gt_SFT": 8.3,
      "pct_RL_gt_Base": 25.0
    },
    {
      "archetype": "coryne_integration",
      "n": 7,
      "Base_mean": -6.7258,
      "SFT_mean": -5.657,
      "RL_mean": -8.5028,
      "RL-SFT": -2.8458,
      "RL-Base": -1.777,
      "median_RL-SFT": -3.822,
      "RL_gt_SFT": 1,
      "RL_gt_Base": 0,
      "pct_RL_gt_SFT": 14.3,
      "pct_RL_gt_Base": 0.0
    },
    {
      "archetype": "mob_F_oriV",
      "n": 8,
      "Base_mean": -5.6372,
      "SFT_mean": -4.9744,
      "RL_mean": -7.338,
      "RL-SFT": -2.3636,
      "RL-Base": -1.7009,
      "median_RL-SFT": -2.8489,
      "RL_gt_SFT": 1,
      "RL_gt_Base": 0,
      "pct_RL_gt_SFT": 12.5,
      "pct_RL_gt_Base": 0.0
    },
    {
      "archetype": "p15A_compact",
      "n": 12,
      "Base_mean": -6.7567,
      "SFT_mean": -5.6706,
      "RL_mean": -7.9451,
      "RL-SFT": -2.2745,
      "RL-Base": -1.1885,
      "median_RL-SFT": -2.2955,
      "RL_gt_SFT": 1,
      "RL_gt_Base": 2,
      "pct_RL_gt_SFT": 8.3,
      "pct_RL_gt_Base": 16.7
    },
    {
      "archetype": "thermus_pIOK",
      "n": 8,
      "Base_mean": -7.4157,
      "SFT_mean": -5.6021,
      "RL_mean": -7.919,
      "RL-SFT": -2.3169,
      "RL-Base": -0.5033,
      "median_RL-SFT": -2.3361,
      "RL_gt_SFT": 0,
      "RL_gt_Base": 1,
      "pct_RL_gt_SFT": 0.0,
      "pct_RL_gt_Base": 12.5
    }
  ],
  "per_archetype_surprisal": [
    {
      "archetype": "BD_patent_sparse",
      "n": 12,
      "Base_mean": -13.6895,
      "SFT_mean": -8.0888,
      "RL_mean": -10.9542,
      "RL-SFT": -2.8654,
      "RL-Base": 2.7353,
      "median_RL-SFT": -3.1824,
      "RL_gt_SFT": 0,
      "RL_gt_Base": 12,
      "pct_RL_gt_SFT": 0.0,
      "pct_RL_gt_Base": 100.0
    },
    {
      "archetype": "coryne_integration",
      "n": 7,
      "Base_mean": -14.0287,
      "SFT_mean": -8.1045,
      "RL_mean": -11.3936,
      "RL-SFT": -3.2891,
      "RL-Base": 2.6351,
      "median_RL-SFT": -2.8462,
      "RL_gt_SFT": 0,
      "RL_gt_Base": 7,
      "pct_RL_gt_SFT": 0.0,
      "pct_RL_gt_Base": 100.0
    },
    {
      "archetype": "mob_F_oriV",
      "n": 8,
      "Base_mean": -12.7867,
      "SFT_mean": -7.3515,
      "RL_mean": -10.207,
      "RL-SFT": -2.8555,
      "RL-Base": 2.5797,
      "median_RL-SFT": -3.0056,
      "RL_gt_SFT": 0,
      "RL_gt_Base": 8,
      "pct_RL_gt_SFT": 0.0,
      "pct_RL_gt_Base": 100.0
    },
    {
      "archetype": "p15A_compact",
      "n": 12,
      "Base_mean": -13.6472,
      "SFT_mean": -8.4667,
      "RL_mean": -11.5608,
      "RL-SFT": -3.0941,
      "RL-Base": 2.0864,
      "median_RL-SFT": -3.5547,
      "RL_gt_SFT": 0,
      "RL_gt_Base": 12,
      "pct_RL_gt_SFT": 0.0,
      "pct_RL_gt_Base": 100.0
    },
    {
      "archetype": "thermus_pIOK",
      "n": 8,
      "Base_mean": -12.9411,
      "SFT_mean": -8.0559,
      "RL_mean": -10.923,
      "RL-SFT": -2.8671,
      "RL-Base": 2.0181,
      "median_RL-SFT": -3.0647,
      "RL_gt_SFT": 0,
      "RL_gt_Base": 8,
      "pct_RL_gt_SFT": 0.0,
      "pct_RL_gt_Base": 100.0
    }
  ],
  "overall": {
    "comp_RL_minus_SFT_mean": -2.3578154256305957,
    "comp_n_RL_gt_SFT": 4,
    "surp_RL_minus_SFT_mean": -2.985494414892844,
    "surp_n_RL_gt_SFT": 0
  },
  "both_metric_wins": 0,
  "comp_only_wins": 4,
  "surp_only_wins": 0
}