{
  "id": 116801,
  "title": "what is the significance of candidate_index in annotations field  and top_level in long_answer_candidates field ",
  "url": "/competitions/tensorflow2-question-answering/discussion/116801",
  "author_name": "",
  "post_date": "2019-11-11T14:15:05.796305600Z",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "",
  "messages": [
    {
      "id": "670497",
      "postDate": "11/11/2019 14:15:05",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "671372",
      "postDate": "11/12/2019 15:27:18",
      "content": "<p>Don't remember accurately but I think it's used to indicate the correct answer.</p>",
      "rawMarkdown": "Don't remember accurately but I think it's used to indicate the correct answer.",
      "votes": null
    },
    {
      "id": "683966",
      "postDate": "11/29/2019 04:53:11",
      "content": "<p>If you see for each observation there are multiple answers under <code>long_answer_candidates</code>. It is a list of possible indexes. So, assume <code>len(entry['long_answer_candidates']) = 481</code>, if <code>candidate_index=59</code>, then <code>correct answer = entry['long_answer_candidates'][59]</code> .</p>",
      "rawMarkdown": "If you see for each observation there are multiple answers under ```long_answer_candidates```. It is a list of possible indexes. So, assume ```len(entry['long_answer_candidates']) = 481```, if ```candidate_index=59```, then ```correct answer = entry['long_answer_candidates'][59]``` .",
      "votes": null
    },
    {
      "id": "685640",
      "postDate": "12/02/2019 03:19:20",
      "content": "<p>Top_level means whether this sentence is contained by other sentences. check this out: <a href=\"https://github.com/google-research-datasets/natural-questions/blob/master/README.md\">https://github.com/google-research-datasets/natural-questions/blob/master/README.md</a></p>",
      "rawMarkdown": "Top_level means whether this sentence is contained by other sentences. check this out: https://github.com/google-research-datasets/natural-questions/blob/master/README.md",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 671372,
      "author_name": "arvindpdmn",
      "author_url": "",
      "post_date": "11/12/2019 15:27:18",
      "content": "<p>Don't remember accurately but I think it's used to indicate the correct answer.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 683966,
      "author_name": "s4sarath",
      "author_url": "",
      "post_date": "11/29/2019 04:53:11",
      "content": "<p>If you see for each observation there are multiple answers under <code>long_answer_candidates</code>. It is a list of possible indexes. So, assume <code>len(entry['long_answer_candidates']) = 481</code>, if <code>candidate_index=59</code>, then <code>correct answer = entry['long_answer_candidates'][59]</code> .</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 685640,
      "author_name": "tuiq007",
      "author_url": "",
      "post_date": "12/02/2019 03:19:20",
      "content": "<p>Top_level means whether this sentence is contained by other sentences. check this out: <a href=\"https://github.com/google-research-datasets/natural-questions/blob/master/README.md\">https://github.com/google-research-datasets/natural-questions/blob/master/README.md</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "670497": "",
    "671372": "Don't remember accurately but I think it's used to indicate the correct answer.",
    "683966": "If you see for each observation there are multiple answers under ```long_answer_candidates```. It is a list of possible indexes. So, assume ```len(entry['long_answer_candidates']) = 481```, if ```candidate_index=59```, then ```correct answer = entry['long_answer_candidates'][59]``` .",
    "685640": "Top_level means whether this sentence is contained by other sentences. check this out: https://github.com/google-research-datasets/natural-questions/blob/master/README.md"
  },
  "source": "meta"
}