sets.json

[
  {
    "set": "synth",
    "n_questions": 6500,
    "n_contexts": 208,
    "corpus_file": "corpus_synth.json",
    "eval_file": "eval_synth.json",
    "facets": [
      {
        "key": "Task group",
        "values": [
          "counting",
          "timeline",
          "user"
        ]
      },
      {
        "key": "Answer type",
        "values": [
          "COMPARISON",
          "DATE",
          "LABEL",
          "MONTH_YEAR",
          "NUMERIC",
          "USER"
        ]
      },
      {
        "key": "Source dataset",
        "values": [
          "agnews",
          "app_reviews",
          "formality",
          "imdb",
          "metaphors",
          "multinli",
          "negation",
          "spam",
          "trec_coarse",
          "yahoo"
        ]
      }
    ]
  },
  {
    "set": "real",
    "n_questions": 949,
    "n_contexts": 22,
    "corpus_file": "corpus_real.json",
    "eval_file": "eval_real.json",
    "facets": [
      {
        "key": "Question type",
        "values": [
          "multidoc_rolls",
          "multidoc_spells",
          "singledoc_rolls",
          "singledoc_spells"
        ]
      },
      {
        "key": "Split",
        "values": [
          "test",
          "validation"
        ]
      },
      {
        "key": "Campaign",
        "values": [
          "campaign1",
          "campaign2"
        ]
      }
    ]
  }
]