{
  "document_id": "lb_qasper_34e4bb298d4e2c19c2fce8670a00196cddd9f361eeb09a9a",
  "family": "public_qasper",
  "split": "dev",
  "document_hash": "964d476624da8b128caa47d71c9c02eb14daae7f3eded545935312e44377ae12",
  "questions": [
    {
      "question_id": "q00",
      "qtype": "qasper",
      "question": "How does Gaussian-masked directional multi-head attention works?",
      "answer": "pays attentions to adjacent characters and casts a localness relationship between the characters as a fixed Gaussian weight assuming the weight relies on the distance between characters",
      "aliases": [
        "Gaussian-masked directional attention can be described as a function to map queries and key-value pairs to the representation of input, Gaussian-masked directional attention expects to pay attention to the adjacent characters of each positions and cast the localness relationship between characters as a fix Gaussian weight for attention, Gaussian weight only relys on the distance between characters"
      ],
      "answer_type": "qa_f1",
      "tolerance": 0.0,
      "supporting_fact_ids": [],
      "answer_position_frac": -1.0
    }
  ],
  "facts": []
}