{
 "n_queries": 15,
 "query_ids": [
  "125",
  "132",
  "276",
  "286",
  "543",
  "614",
  "695",
  "713",
  "835",
  "915",
  "919",
  "926",
  "936",
  "1106",
  "1126"
 ],
 "selection_method": "stratified_3",
 "n_runs": 79,
 "n_systems": 18,
 "split_systems": {
  "select": [
   "timchen0618 (one LLM, 6 trajectory-summary variants)",
   "OpenResearcher-30B-A3B @ OpenResearcher eval logs",
   "Qwen3.5-35B-A3B @ liuqi6777 search-agent trajectories",
   "Claude-Opus-4.7 @ Pi-Serini",
   "DeepSeek-V4-Pro @ liuqi6777 search-agent trajectories",
   "Kimi-K2.6 @ liuqi6777 search-agent trajectories"
  ],
  "validate": [
   "Tongyi-DeepResearch-30B-A3B @ liuqi6777 search-agent trajectories",
   "Claude-Haiku-4.5 @ Pi-Serini",
   "DeepSeek-V4-Pro @ Pi-Serini",
   "GPT-5 @ BrowseComp-Plus official runs",
   "GPT-5.4 @ Pi-Serini",
   "GPT-5.5 @ Pi-Serini"
  ],
  "test": [
   "gpt-oss-120b-high @ liuqi6777 search-agent trajectories",
   "gpt-oss-20b (high) @ Revisiting Text Ranking in Deep Research (SIGIR'26) runs",
   "o3 @ BrowseComp-Plus official runs",
   "DeepSeek-V4-Flash @ Pi-Serini",
   "GPT-5.2 @ Pi-Serini",
   "GPT-5 @ Pi-Serini"
  ]
 },
 "test_split_weighted_spearman": 0.9734,
 "test_split_mae_after_linear_fit_pts": 3.75,
 "test_split_raw_mae_pts": 14.99,
 "random_subset_test_split_weighted_spearman": 0.8055,
 "all_runs_weighted_spearman": 0.9503,
 "random_subset_all_runs_weighted_spearman": 0.7456,
 "all_runs_mae_after_linear_fit_pts": 2.93,
 "linear_fit_full_from_mini": {
  "slope": 0.3076,
  "intercept": 0.4279
 },
 "full_benchmark_split_half_weighted_spearman": 0.9741,
 "size_curve_mean_over_splits": {
  "10": {
   "test_spearman_mean": 0.8622,
   "test_spearman_sd": 0.0845,
   "random_test_spearman_mean": 0.6817,
   "test_calibrated_mae_pts": 5.42
  },
  "12": {
   "test_spearman_mean": 0.886,
   "test_spearman_sd": 0.1015,
   "random_test_spearman_mean": 0.7146,
   "test_calibrated_mae_pts": 4.99
  },
  "15": {
   "test_spearman_mean": 0.8779,
   "test_spearman_sd": 0.0868,
   "random_test_spearman_mean": 0.7492,
   "test_calibrated_mae_pts": 5.08
  },
  "20": {
   "test_spearman_mean": 0.8869,
   "test_spearman_sd": 0.0705,
   "random_test_spearman_mean": 0.7979,
   "test_calibrated_mae_pts": 5.44
  },
  "30": {
   "test_spearman_mean": 0.8955,
   "test_spearman_sd": 0.0738,
   "random_test_spearman_mean": 0.8467,
   "test_calibrated_mae_pts": 4.75
  }
 }
}