data/sft/split_audit.json

{
  "source": "data/eval_full_10985.jsonl",
  "source_sha256": "e898872e0ee12d177c988bde6f69687fb4761c20bafd4f74ff64af40150e0b90",
  "seed": 20260817,
  "group_key": "case-folded stripped query",
  "stratification": [
    "strategy_id",
    "query_bucket",
    "complexified"
  ],
  "requested_fractions": {
    "validation": 0.05,
    "test": 0.1
  },
  "rows": {
    "train": 9341,
    "validation": 549,
    "test": 1095
  },
  "unique_queries": {
    "train": 9333,
    "validation": 548,
    "test": 1094
  },
  "disjoint_query_groups": true,
  "by_split": {
    "train": {
      "strategies": {
        "browsing_inspiration": 1627,
        "directed_detailed": 1642,
        "general": 3276,
        "negation_complex_expressions": 1210,
        "value_deals": 1586
      },
      "query_buckets": {
        "1-10": 1547,
        "101+": 1087,
        "11-30": 2021,
        "31-60": 2410,
        "61-100": 2276
      },
      "complexified": 2808,
      "has_null": 218,
      "has_empty_array": 35,
      "has_nested_object": 4836
    },
    "validation": {
      "strategies": {
        "browsing_inspiration": 94,
        "directed_detailed": 98,
        "general": 193,
        "negation_complex_expressions": 71,
        "value_deals": 93
      },
      "query_buckets": {
        "1-10": 92,
        "101+": 63,
        "11-30": 120,
        "31-60": 141,
        "61-100": 133
      },
      "complexified": 164,
      "has_null": 11,
      "has_empty_array": 3,
      "has_nested_object": 271
    },
    "test": {
      "strategies": {
        "browsing_inspiration": 190,
        "directed_detailed": 191,
        "general": 384,
        "negation_complex_expressions": 142,
        "value_deals": 188
      },
      "query_buckets": {
        "1-10": 182,
        "101+": 127,
        "11-30": 237,
        "31-60": 281,
        "61-100": 268
      },
      "complexified": 328,
      "has_null": 31,
      "has_empty_array": 2,
      "has_nested_object": 554
    }
  }
}