{
  "id": 67139,
  "title": "Stage 2 questions",
  "url": "/competitions/inclusive-images-challenge/discussion/67139",
  "author_name": "",
  "post_date": "2018-09-29T08:55:45.729011100Z",
  "votes": 7,
  "comment_count": 1,
  "views": 0,
  "content": "<ol>\n<li><p>Can we expect that 2nd stage has approximately same distribution\nof classes as in stage1 test set?</p></li>\n<li><p>Can we expect that test set in 2nd stage was made by the same\nannotators with same annotation rules as in 1st stage? Was it\nannotated in the same time?</p></li>\n<li><p>Can we upload two models as in other competitions? It would be\ngood since there are too many unknown factors for stage 2 test set.\nI'd like to upload \"risky\" model as well.</p></li>\n</ol>",
  "messages": [
    {
      "id": "395746",
      "postDate": "09/29/2018 08:55:45",
      "content": "<ol>\n<li><p>Can we expect that 2nd stage has approximately same distribution\nof classes as in stage1 test set?</p></li>\n<li><p>Can we expect that test set in 2nd stage was made by the same\nannotators with same annotation rules as in 1st stage? Was it\nannotated in the same time?</p></li>\n<li><p>Can we upload two models as in other competitions? It would be\ngood since there are too many unknown factors for stage 2 test set.\nI'd like to upload \"risky\" model as well.</p></li>\n</ol>",
      "rawMarkdown": "1. Can we expect that 2nd stage has approximately same distribution\n   of classes as in stage1 test set?\n   \n   2. Can we expect that test set in 2nd stage was made by the same\n   annotators with same annotation rules as in 1st stage? Was it\n   annotated in the same time?\n   \n   3. Can we upload two models as in other competitions? It would be\n   good since there are too many unknown factors for stage 2 test set.\n   I'd like to upload \"risky\" model as well.",
      "votes": null
    },
    {
      "id": "396854",
      "postDate": "10/01/2018 13:38:22",
      "content": "<p>I think the answer to your first and partly to the second questions are in the FAQ. Answer to 1st is No, they are likely to be quite different. Answer to 2 (1) is yes. </p>",
      "rawMarkdown": "I think the answer to your first and partly to the second questions are in the FAQ. Answer to 1st is No, they are likely to be quite different. Answer to 2 (1) is yes.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 396854,
      "author_name": "amrit110",
      "author_url": "",
      "post_date": "10/01/2018 13:38:22",
      "content": "<p>I think the answer to your first and partly to the second questions are in the FAQ. Answer to 1st is No, they are likely to be quite different. Answer to 2 (1) is yes. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "395746": "1. Can we expect that 2nd stage has approximately same distribution\n   of classes as in stage1 test set?\n   \n   2. Can we expect that test set in 2nd stage was made by the same\n   annotators with same annotation rules as in 1st stage? Was it\n   annotated in the same time?\n   \n   3. Can we upload two models as in other competitions? It would be\n   good since there are too many unknown factors for stage 2 test set.\n   I'd like to upload \"risky\" model as well.",
    "396854": "I think the answer to your first and partly to the second questions are in the FAQ. Answer to 1st is No, they are likely to be quite different. Answer to 2 (1) is yes."
  },
  "source": "meta"
}