{
  "id": 52053,
  "title": "> 1 hour ? ",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/52053",
  "author_name": "",
  "post_date": "2018-03-15T14:21:29.014299400Z",
  "votes": 1,
  "comment_count": 6,
  "views": 0,
  "content": "<p>So now kernel can be up to more than one hour?</p>",
  "messages": [
    {
      "id": "296614",
      "postDate": "03/15/2018 14:21:29",
      "content": "<p>So now kernel can be up to more than one hour?</p>",
      "rawMarkdown": "So now kernel can be up to more than one hour?",
      "votes": null
    },
    {
      "id": "296629",
      "postDate": "03/15/2018 14:40:04",
      "content": "<p>Seems to be 4 hrs.....</p>",
      "rawMarkdown": "Seems to be 4 hrs.....",
      "votes": null
    },
    {
      "id": "296757",
      "postDate": "03/15/2018 18:06:50",
      "content": "<p>6 hours at most</p>",
      "rawMarkdown": "6 hours at most",
      "votes": null
    },
    {
      "id": "296926",
      "postDate": "03/16/2018 01:24:48",
      "content": "<p>And we can chain a lot of them, seems kaggle really want to push kernel this year.</p>\n\n<p>Kernel based solution might be better for costumer I think. </p>",
      "rawMarkdown": "And we can chain a lot of them, seems kaggle really want to push kernel this year.\n\nKernel based solution might be better for costumer I think.",
      "votes": null
    },
    {
      "id": "296930",
      "postDate": "03/16/2018 01:42:27",
      "content": "<p>I need more memory :(</p>",
      "rawMarkdown": "I need more memory :(",
      "votes": null
    },
    {
      "id": "296932",
      "postDate": "03/16/2018 01:52:05",
      "content": "<p>How about bagged models? I see even 40000000 only data can give a very good score.</p>",
      "rawMarkdown": "How about bagged models? I see even 40000000 only data can give a very good score.",
      "votes": null
    },
    {
      "id": "297000",
      "postDate": "03/16/2018 04:35:46",
      "content": "<p>Still working on Toxic. Sounds like a good idea!</p>",
      "rawMarkdown": "Still working on Toxic. Sounds like a good idea!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 296629,
      "author_name": "shujian",
      "author_url": "",
      "post_date": "03/15/2018 14:40:04",
      "content": "<p>Seems to be 4 hrs.....</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 296757,
      "author_name": "konohayui",
      "author_url": "",
      "post_date": "03/15/2018 18:06:50",
      "content": "<p>6 hours at most</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 296926,
      "author_name": "muhammadalfiansyah",
      "author_url": "",
      "post_date": "03/16/2018 01:24:48",
      "content": "<p>And we can chain a lot of them, seems kaggle really want to push kernel this year.</p>\n\n<p>Kernel based solution might be better for costumer I think. </p>",
      "votes": null,
      "replies": [
        {
          "id": 296930,
          "author_name": "shujian",
          "author_url": "",
          "post_date": "03/16/2018 01:42:27",
          "content": "<p>I need more memory :(</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 296932,
          "author_name": "muhammadalfiansyah",
          "author_url": "",
          "post_date": "03/16/2018 01:52:05",
          "content": "<p>How about bagged models? I see even 40000000 only data can give a very good score.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 297000,
          "author_name": "shujian",
          "author_url": "",
          "post_date": "03/16/2018 04:35:46",
          "content": "<p>Still working on Toxic. Sounds like a good idea!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "296614": "So now kernel can be up to more than one hour?",
    "296629": "Seems to be 4 hrs.....",
    "296757": "6 hours at most",
    "296926": "And we can chain a lot of them, seems kaggle really want to push kernel this year.\n\nKernel based solution might be better for costumer I think.",
    "296930": "I need more memory :(",
    "296932": "How about bagged models? I see even 40000000 only data can give a very good score.",
    "297000": "Still working on Toxic. Sounds like a good idea!"
  },
  "source": "meta"
}