{
  "id": 76442,
  "title": "Saving files across Kernel restart",
  "url": "/competitions/quora-insincere-questions-classification/discussion/76442",
  "author_name": "",
  "post_date": "2019-01-02T21:45:21.095611200Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>To speed up my development time, I'd like to store some files (not externally sourced) and load them for use.  These files will also come in handy for faster training.</p>\n\n<ol>\n<li>How can I keep the generated files across kernel restarts? </li>\n<li>If we are to submit the Kernel code, does that mean that the sponsor will first run the training, and then, run the prediction, rather than a trained model?  In other words, does the 2-hour/6-hour time limit include both training and prediction?</li>\n</ol>\n\n<p>(This is my first Kaggle project and first competition.)</p>",
  "messages": [
    {
      "id": "449255",
      "postDate": "01/02/2019 21:45:21",
      "content": "<p>To speed up my development time, I'd like to store some files (not externally sourced) and load them for use.  These files will also come in handy for faster training.</p>\n\n<ol>\n<li>How can I keep the generated files across kernel restarts? </li>\n<li>If we are to submit the Kernel code, does that mean that the sponsor will first run the training, and then, run the prediction, rather than a trained model?  In other words, does the 2-hour/6-hour time limit include both training and prediction?</li>\n</ol>\n\n<p>(This is my first Kaggle project and first competition.)</p>",
      "rawMarkdown": "To speed up my development time, I'd like to store some files (not externally sourced) and load them for use.  These files will also come in handy for faster training.\n\n1. How can I keep the generated files across kernel restarts? \n2. If we are to submit the Kernel code, does that mean that the sponsor will first run the training, and then, run the prediction, rather than a trained model?  In other words, does the 2-hour/6-hour time limit include both training and prediction?\n\n(This is my first Kaggle project and first competition.)",
      "votes": null
    },
    {
      "id": "449270",
      "postDate": "01/02/2019 22:10:41",
      "content": "<ol>\n<li>You cannot automatically afaik. Only if you download and reupload as dataset.</li>\n<li>Yes, limit includes both training and prediction.</li>\n</ol>",
      "rawMarkdown": "1. You cannot automatically afaik. Only if you download and reupload as dataset.\n2. Yes, limit includes both training and prediction.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 449270,
      "author_name": "philippsinger",
      "author_url": "",
      "post_date": "01/02/2019 22:10:41",
      "content": "<ol>\n<li>You cannot automatically afaik. Only if you download and reupload as dataset.</li>\n<li>Yes, limit includes both training and prediction.</li>\n</ol>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "449255": "To speed up my development time, I'd like to store some files (not externally sourced) and load them for use.  These files will also come in handy for faster training.\n\n1. How can I keep the generated files across kernel restarts? \n2. If we are to submit the Kernel code, does that mean that the sponsor will first run the training, and then, run the prediction, rather than a trained model?  In other words, does the 2-hour/6-hour time limit include both training and prediction?\n\n(This is my first Kaggle project and first competition.)",
    "449270": "1. You cannot automatically afaik. Only if you download and reupload as dataset.\n2. Yes, limit includes both training and prediction."
  },
  "source": "meta"
}