{
  "id": 95225,
  "title": "Feedback Collection Thread",
  "url": "/competitions/imet-2019-fgvc6/discussion/95225",
  "author_name": "",
  "post_date": "2019-06-10T22:24:21.344587500Z",
  "votes": 5,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi Dear Fellow Masters (and soon-to-be's),</p>\n\n<p>Thanks again for the effort and time you have spent in the competition! To best recap this year's competition and prepare for the next one, we are collection information about your feelings during the last 2 months:\n<a href=\"https://forms.gle/gRi3e2UZ9pWvxNVy8\">Please fill out this form</a></p>\n\n<p>(Please also use the link if you are willing to visit and/or present in our workshop FGVC6 at <a href=\"https://sites.google.com/corp/view/fgvc6/home\">CVPR'19</a>)</p>\n\n<p>Looking forward to your responses!</p>",
  "messages": [
    {
      "id": "549634",
      "postDate": "06/10/2019 22:24:21",
      "content": "<p>Hi Dear Fellow Masters (and soon-to-be's),</p>\n\n<p>Thanks again for the effort and time you have spent in the competition! To best recap this year's competition and prepare for the next one, we are collection information about your feelings during the last 2 months:\n<a href=\"https://forms.gle/gRi3e2UZ9pWvxNVy8\">Please fill out this form</a></p>\n\n<p>(Please also use the link if you are willing to visit and/or present in our workshop FGVC6 at <a href=\"https://sites.google.com/corp/view/fgvc6/home\">CVPR'19</a>)</p>\n\n<p>Looking forward to your responses!</p>",
      "rawMarkdown": "Hi Dear Fellow Masters (and soon-to-be's),\n\nThanks again for the effort and time you have spent in the competition! To best recap this year's competition and prepare for the next one, we are collection information about your feelings during the last 2 months:\n[Please fill out this form](https://forms.gle/gRi3e2UZ9pWvxNVy8)\n\n(Please also use the link if you are willing to visit and/or present in our workshop FGVC6 at [CVPR'19](https://sites.google.com/corp/view/fgvc6/home))\n\nLooking forward to your responses!",
      "votes": null
    },
    {
      "id": "549697",
      "postDate": "06/11/2019 00:10:11",
      "content": "<p>Can we do late submission after competition end?</p>",
      "rawMarkdown": "Can we do late submission after competition end?",
      "votes": null
    },
    {
      "id": "550467",
      "postDate": "06/11/2019 16:47:51",
      "content": "<p>Stay tuned:)</p>",
      "rawMarkdown": "Stay tuned:)",
      "votes": null
    },
    {
      "id": "550717",
      "postDate": "06/12/2019 00:51:17",
      "content": "<p>Thank you for hosting interesting competition. I really enjoyed it!</p>\n\n<p>But,,, maybe same as other competitors, I feel gloomy because of the mess at the end of competition...\nIn <a href=\"https://www.kaggle.com/c/imet-2019-fgvc6/discussion/94430#latest-550445\">this thread</a>, <a href=\"/dempton\">@dempton</a> said top 5 in the public LB scrapped the labels for public test set from web. It seems somehow true because their scores were incredibly high and some of them were drastically dropped at 2nd stage. That fact supports what <a href=\"/dempton\">@dempton</a> said, doesn't it?\nAnd, some of them still remains in public LB. It seems they were not guilty, but is it true? If someone can get labels for 10K test images, that means they have additional 10K training set. It's big advantage to other competitors.</p>\n\n<p>I'm not saying they are cheater. If they got such high score without anything wrong, I really want to admire them. And I assume they also want to make their doubt clear. (Who wants to called cheater?) Some says they should open their codes, but I can understand if it's truly good they don't want to reveal it.</p>\n\n<p>So, I want to ask you whether you have already reviewed the reproducibility of codes from someone who got high score, at least in gold zone?\nI definitely believe if you kindly answer the question, it will make everyone feel clear and comfortable.</p>\n\n<p>Thanks, I'm looking for the next one!</p>",
      "rawMarkdown": "Thank you for hosting interesting competition. I really enjoyed it!\n\nBut,,, maybe same as other competitors, I feel gloomy because of the mess at the end of competition...\nIn [this thread](https://www.kaggle.com/c/imet-2019-fgvc6/discussion/94430#latest-550445), @dempton said top 5 in the public LB scrapped the labels for public test set from web. It seems somehow true because their scores were incredibly high and some of them were drastically dropped at 2nd stage. That fact supports what @dempton said, doesn't it?\nAnd, some of them still remains in public LB. It seems they were not guilty, but is it true? If someone can get labels for 10K test images, that means they have additional 10K training set. It's big advantage to other competitors.\n\nI'm not saying they are cheater. If they got such high score without anything wrong, I really want to admire them. And I assume they also want to make their doubt clear. (Who wants to called cheater?) Some says they should open their codes, but I can understand if it's truly good they don't want to reveal it.\n\nSo, I want to ask you whether you have already reviewed the reproducibility of codes from someone who got high score, at least in gold zone?\nI definitely believe if you kindly answer the question, it will make everyone feel clear and comfortable.\n\nThanks, I'm looking for the next one!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 549697,
      "author_name": "soywugzm",
      "author_url": "",
      "post_date": "06/11/2019 00:10:11",
      "content": "<p>Can we do late submission after competition end?</p>",
      "votes": null,
      "replies": [
        {
          "id": 550467,
          "author_name": "codingcliff",
          "author_url": "",
          "post_date": "06/11/2019 16:47:51",
          "content": "<p>Stay tuned:)</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 550717,
      "author_name": "bamps53",
      "author_url": "",
      "post_date": "06/12/2019 00:51:17",
      "content": "<p>Thank you for hosting interesting competition. I really enjoyed it!</p>\n\n<p>But,,, maybe same as other competitors, I feel gloomy because of the mess at the end of competition...\nIn <a href=\"https://www.kaggle.com/c/imet-2019-fgvc6/discussion/94430#latest-550445\">this thread</a>, <a href=\"/dempton\">@dempton</a> said top 5 in the public LB scrapped the labels for public test set from web. It seems somehow true because their scores were incredibly high and some of them were drastically dropped at 2nd stage. That fact supports what <a href=\"/dempton\">@dempton</a> said, doesn't it?\nAnd, some of them still remains in public LB. It seems they were not guilty, but is it true? If someone can get labels for 10K test images, that means they have additional 10K training set. It's big advantage to other competitors.</p>\n\n<p>I'm not saying they are cheater. If they got such high score without anything wrong, I really want to admire them. And I assume they also want to make their doubt clear. (Who wants to called cheater?) Some says they should open their codes, but I can understand if it's truly good they don't want to reveal it.</p>\n\n<p>So, I want to ask you whether you have already reviewed the reproducibility of codes from someone who got high score, at least in gold zone?\nI definitely believe if you kindly answer the question, it will make everyone feel clear and comfortable.</p>\n\n<p>Thanks, I'm looking for the next one!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "549634": "Hi Dear Fellow Masters (and soon-to-be's),\n\nThanks again for the effort and time you have spent in the competition! To best recap this year's competition and prepare for the next one, we are collection information about your feelings during the last 2 months:\n[Please fill out this form](https://forms.gle/gRi3e2UZ9pWvxNVy8)\n\n(Please also use the link if you are willing to visit and/or present in our workshop FGVC6 at [CVPR'19](https://sites.google.com/corp/view/fgvc6/home))\n\nLooking forward to your responses!",
    "549697": "Can we do late submission after competition end?",
    "550467": "Stay tuned:)",
    "550717": "Thank you for hosting interesting competition. I really enjoyed it!\n\nBut,,, maybe same as other competitors, I feel gloomy because of the mess at the end of competition...\nIn [this thread](https://www.kaggle.com/c/imet-2019-fgvc6/discussion/94430#latest-550445), @dempton said top 5 in the public LB scrapped the labels for public test set from web. It seems somehow true because their scores were incredibly high and some of them were drastically dropped at 2nd stage. That fact supports what @dempton said, doesn't it?\nAnd, some of them still remains in public LB. It seems they were not guilty, but is it true? If someone can get labels for 10K test images, that means they have additional 10K training set. It's big advantage to other competitors.\n\nI'm not saying they are cheater. If they got such high score without anything wrong, I really want to admire them. And I assume they also want to make their doubt clear. (Who wants to called cheater?) Some says they should open their codes, but I can understand if it's truly good they don't want to reveal it.\n\nSo, I want to ask you whether you have already reviewed the reproducibility of codes from someone who got high score, at least in gold zone?\nI definitely believe if you kindly answer the question, it will make everyone feel clear and comfortable.\n\nThanks, I'm looking for the next one!"
  },
  "source": "meta"
}