{
  "id": 272501,
  "title": "IMPORTANT: Updated Dataset",
  "url": "/competitions/wikipedia-image-caption/discussion/272501",
  "author_name": "",
  "post_date": "2021-09-15T19:10:40.946862100Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello everyone -</p>\n<p>There has been two adjustments to the dataset:</p>\n<ol>\n<li>The <code>test.tsv</code> file only contains <code>id</code> and <code>image_url</code> columns. <em>Please</em> only use this updated file, and <em>do not</em> publish the previous version. (Thanks in advance for that!)</li>\n<li>A <code>test_caption_list.csv</code> file is included in the dataset that includes a list of captions to be retrieved for predictions on the test data.</li>\n</ol>",
  "messages": [
    {
      "id": "1514199",
      "postDate": "09/15/2021 19:10:40",
      "content": "<p>Hello everyone -</p>\n<p>There has been two adjustments to the dataset:</p>\n<ol>\n<li>The <code>test.tsv</code> file only contains <code>id</code> and <code>image_url</code> columns. <em>Please</em> only use this updated file, and <em>do not</em> publish the previous version. (Thanks in advance for that!)</li>\n<li>A <code>test_caption_list.csv</code> file is included in the dataset that includes a list of captions to be retrieved for predictions on the test data.</li>\n</ol>",
      "rawMarkdown": "Hello everyone -\n\nThere has been two adjustments to the dataset:\n\n1. The `test.tsv` file only contains `id` and `image_url` columns. _Please_ only use this updated file, and _do not_ publish the previous version. (Thanks in advance for that!)\n2. A `test_caption_list.csv` file is included in the dataset that includes a list of captions to be retrieved for predictions on the test data.",
      "votes": null
    },
    {
      "id": "1514463",
      "postDate": "09/16/2021 05:28:33",
      "content": "<p>Thanks for the update <a href=\"https://www.kaggle.com/inversion\" target=\"_blank\">@inversion</a>, I was boggled about the missing test caption list before :) How about language now - should we pick captions closest to the image independent of language they are written in? </p>\n<p>Also, there is potential leakage in the url link. Is this acceptable to use it, or are we being asked to neglect that url and just focus on the images? </p>",
      "rawMarkdown": "Thanks for the update @inversion, I was boggled about the missing test caption list before :) How about language now - should we pick captions closest to the image independent of language they are written in? \n\nAlso, there is potential leakage in the url link. Is this acceptable to use it, or are we being asked to neglect that url and just focus on the images?",
      "votes": null
    },
    {
      "id": "1612491",
      "postDate": "12/09/2021 02:49:06",
      "content": "<p>\"All deadlines are at 11:59 <strong>PM</strong> UTC on the corresponding day unless otherwise noted. \"<br>\nBut why the date displayed is 11:59 <strong>AM</strong> UTC? <a href=\"https://www.kaggle.com/inversion\" target=\"_blank\">@inversion</a> </p>",
      "rawMarkdown": "\"All deadlines are at 11:59 **PM** UTC on the corresponding day unless otherwise noted. \"\nBut why the date displayed is 11:59 **AM** UTC? @inversion",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1514463,
      "author_name": "thedrcat",
      "author_url": "",
      "post_date": "09/16/2021 05:28:33",
      "content": "<p>Thanks for the update <a href=\"https://www.kaggle.com/inversion\" target=\"_blank\">@inversion</a>, I was boggled about the missing test caption list before :) How about language now - should we pick captions closest to the image independent of language they are written in? </p>\n<p>Also, there is potential leakage in the url link. Is this acceptable to use it, or are we being asked to neglect that url and just focus on the images? </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1612491,
      "author_name": "whutzy",
      "author_url": "",
      "post_date": "12/09/2021 02:49:06",
      "content": "<p>\"All deadlines are at 11:59 <strong>PM</strong> UTC on the corresponding day unless otherwise noted. \"<br>\nBut why the date displayed is 11:59 <strong>AM</strong> UTC? <a href=\"https://www.kaggle.com/inversion\" target=\"_blank\">@inversion</a> </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1514199": "Hello everyone -\n\nThere has been two adjustments to the dataset:\n\n1. The `test.tsv` file only contains `id` and `image_url` columns. _Please_ only use this updated file, and _do not_ publish the previous version. (Thanks in advance for that!)\n2. A `test_caption_list.csv` file is included in the dataset that includes a list of captions to be retrieved for predictions on the test data.",
    "1514463": "Thanks for the update @inversion, I was boggled about the missing test caption list before :) How about language now - should we pick captions closest to the image independent of language they are written in? \n\nAlso, there is potential leakage in the url link. Is this acceptable to use it, or are we being asked to neglect that url and just focus on the images?",
    "1612491": "\"All deadlines are at 11:59 **PM** UTC on the corresponding day unless otherwise noted. \"\nBut why the date displayed is 11:59 **AM** UTC? @inversion"
  },
  "source": "meta"
}