{
  "id": 73310,
  "title": "Can I train on my local computer?",
  "url": "/competitions/quora-insincere-questions-classification/discussion/73310",
  "author_name": "",
  "post_date": "2018-12-01T20:16:38.405032500Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Sorry, I am kind of new to Kaggle. For this competition can I train on my local computer or I need to use Kernels?</p>",
  "messages": [
    {
      "id": "431201",
      "postDate": "12/01/2018 20:16:38",
      "content": "<p>Sorry, I am kind of new to Kaggle. For this competition can I train on my local computer or I need to use Kernels?</p>",
      "rawMarkdown": "Sorry, I am kind of new to Kaggle. For this competition can I train on my local computer or I need to use Kernels?",
      "votes": null
    },
    {
      "id": "431208",
      "postDate": "12/01/2018 20:57:56",
      "content": "<p>It is a kernel only so you can submit only with Kaggle kernels. You can't import external data from your computer either, everything has to be done on your kernel. \nHowever you can do anything you want if you don't want to submit your results.\nHope it helps!</p>",
      "rawMarkdown": "It is a kernel only so you can submit only with Kaggle kernels. You can't import external data from your computer either, everything has to be done on your kernel. \nHowever you can do anything you want if you don't want to submit your results.\nHope it helps!",
      "votes": null
    },
    {
      "id": "431330",
      "postDate": "12/02/2018 04:08:09",
      "content": "<p>Thank you!</p>",
      "rawMarkdown": "Thank you!",
      "votes": null
    },
    {
      "id": "431373",
      "postDate": "12/02/2018 06:00:57",
      "content": "<p>You can of course run lots of tests on your local PC and eventually upload into a kernel and run it there for the official score.  For example, I might run a model that takes 24 hours and use local folding or scoring to decide I have something good.  Then I would need to figure out how to get it to run in Kernel within time limit.  Almost always there is stuff your doing for information - ie, plots of fit history.  Those don't need to be in the final kernel.\nWhile the kernels are a huge improvement over two years ago when I started, I still find them somewhat of a pain to use compared to my home PC.</p>",
      "rawMarkdown": "You can of course run lots of tests on your local PC and eventually upload into a kernel and run it there for the official score.  For example, I might run a model that takes 24 hours and use local folding or scoring to decide I have something good.  Then I would need to figure out how to get it to run in Kernel within time limit.  Almost always there is stuff your doing for information - ie, plots of fit history.  Those don't need to be in the final kernel.\nWhile the kernels are a huge improvement over two years ago when I started, I still find them somewhat of a pain to use compared to my home PC.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 431208,
      "author_name": "theoviel",
      "author_url": "",
      "post_date": "12/01/2018 20:57:56",
      "content": "<p>It is a kernel only so you can submit only with Kaggle kernels. You can't import external data from your computer either, everything has to be done on your kernel. \nHowever you can do anything you want if you don't want to submit your results.\nHope it helps!</p>",
      "votes": null,
      "replies": [
        {
          "id": 431373,
          "author_name": "pcjimmmy",
          "author_url": "",
          "post_date": "12/02/2018 06:00:57",
          "content": "<p>You can of course run lots of tests on your local PC and eventually upload into a kernel and run it there for the official score.  For example, I might run a model that takes 24 hours and use local folding or scoring to decide I have something good.  Then I would need to figure out how to get it to run in Kernel within time limit.  Almost always there is stuff your doing for information - ie, plots of fit history.  Those don't need to be in the final kernel.\nWhile the kernels are a huge improvement over two years ago when I started, I still find them somewhat of a pain to use compared to my home PC.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 431330,
      "author_name": "nectario",
      "author_url": "",
      "post_date": "12/02/2018 04:08:09",
      "content": "<p>Thank you!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "431201": "Sorry, I am kind of new to Kaggle. For this competition can I train on my local computer or I need to use Kernels?",
    "431208": "It is a kernel only so you can submit only with Kaggle kernels. You can't import external data from your computer either, everything has to be done on your kernel. \nHowever you can do anything you want if you don't want to submit your results.\nHope it helps!",
    "431330": "Thank you!",
    "431373": "You can of course run lots of tests on your local PC and eventually upload into a kernel and run it there for the official score.  For example, I might run a model that takes 24 hours and use local folding or scoring to decide I have something good.  Then I would need to figure out how to get it to run in Kernel within time limit.  Almost always there is stuff your doing for information - ie, plots of fit history.  Those don't need to be in the final kernel.\nWhile the kernels are a huge improvement over two years ago when I started, I still find them somewhat of a pain to use compared to my home PC."
  },
  "source": "meta"
}