inputs/AgentSelect/other/data_statistic/Query_Diversity/summary.json

{
  "num_queries": 111179,
  "num_sources": 1,
  "parts": {
    "PartI": 23073,
    "PartII": 76197,
    "PartIII": 11909
  },
  "sources_top20": [
    [
      "unknown",
      111179
    ]
  ],
  "global_word_len": {
    "count": 111179,
    "mean": 109.04913697730687,
    "std": 271.09811111390684,
    "min": 0.0,
    "p25": 50.0,
    "median": 78.0,
    "p75": 126.0,
    "max": 45839.0
  },
  "global_lexical_diversity": {
    "count": 111179,
    "mean": 0.6983877802451874,
    "std": 0.14859708818953538,
    "min": 0.0,
    "p25": 0.6062176165803109,
    "median": 0.6986301369863014,
    "p75": 0.8059701492537313,
    "max": 1.0
  },
  "num_exact_duplicate_groups": 12530,
  "num_exact_duplicate_queries": 26921,
  "num_template_groups": 5137,
  "num_near_duplicate_pairs": 106764,
  "dup_ratio_threshold": 0.92,
  "jaccard_ratio_threshold": 0.85,
  "prefix_k": 6
}