{
  "id": 121707,
  "title": "Anyone else running into \"Notebook Exceeded Allowed Compute\"",
  "url": "/competitions/deepfake-detection-challenge/discussion/121707",
  "author_name": "",
  "post_date": "2019-12-15T00:23:50.719058500Z",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I made two submissions today with kernels that are correctly committed, and with enough room in terms of running time and RAM usage. Specifically, I made sure to save generated images in a temporary folder (/kaggle/tmp, so it's not in the output), so that i can run predictions directly by loading mini batches from disk.</p>\n\n<p>However, I ran into \"Notebook Exceeded Allowed Compute\" errors in both cases. Is that likely to be a RAM usage problem, a disk usage problem, or anything else relating to the size of the input?</p>\n\n<p>Would love to have your thoughts if you ran into this problem or solved it for this competition!</p>",
  "messages": [
    {
      "id": "695323",
      "postDate": "12/15/2019 00:23:50",
      "content": "<p>I made two submissions today with kernels that are correctly committed, and with enough room in terms of running time and RAM usage. Specifically, I made sure to save generated images in a temporary folder (/kaggle/tmp, so it's not in the output), so that i can run predictions directly by loading mini batches from disk.</p>\n\n<p>However, I ran into \"Notebook Exceeded Allowed Compute\" errors in both cases. Is that likely to be a RAM usage problem, a disk usage problem, or anything else relating to the size of the input?</p>\n\n<p>Would love to have your thoughts if you ran into this problem or solved it for this competition!</p>",
      "rawMarkdown": "I made two submissions today with kernels that are correctly committed, and with enough room in terms of running time and RAM usage. Specifically, I made sure to save generated images in a temporary folder (/kaggle/tmp, so it's not in the output), so that i can run predictions directly by loading mini batches from disk.\n\nHowever, I ran into \"Notebook Exceeded Allowed Compute\" errors in both cases. Is that likely to be a RAM usage problem, a disk usage problem, or anything else relating to the size of the input?\n\nWould love to have your thoughts if you ran into this problem or solved it for this competition!",
      "votes": null
    },
    {
      "id": "695338",
      "postDate": "12/15/2019 01:26:13",
      "content": "<p>Have not run into compute problems but be careful because the public test set which runs when we submit is much larger than the public validation set we have been given</p>",
      "rawMarkdown": "Have not run into compute problems but be careful because the public test set which runs when we submit is much larger than the public validation set we have been given",
      "votes": null
    },
    {
      "id": "695385",
      "postDate": "12/15/2019 03:32:29",
      "content": "<p>What's the size ratio? Should be less at most 10x no?</p>",
      "rawMarkdown": "What's the size ratio? Should be less at most 10x no?",
      "votes": null
    },
    {
      "id": "695393",
      "postDate": "12/15/2019 03:59:38",
      "content": "<p>I havent directly measured but my first commit seems to take 3-5 minutes and my submit does seem to be about 10x that time. </p>",
      "rawMarkdown": "I havent directly measured but my first commit seems to take 3-5 minutes and my submit does seem to be about 10x that time.",
      "votes": null
    },
    {
      "id": "695887",
      "postDate": "12/15/2019 18:57:22",
      "content": "<p>my evaluation is also 10x</p>",
      "rawMarkdown": "my evaluation is also 10x",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 695338,
      "author_name": "ryches",
      "author_url": "",
      "post_date": "12/15/2019 01:26:13",
      "content": "<p>Have not run into compute problems but be careful because the public test set which runs when we submit is much larger than the public validation set we have been given</p>",
      "votes": null,
      "replies": [
        {
          "id": 695385,
          "author_name": "xhlulu",
          "author_url": "",
          "post_date": "12/15/2019 03:32:29",
          "content": "<p>What's the size ratio? Should be less at most 10x no?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 695393,
          "author_name": "ryches",
          "author_url": "",
          "post_date": "12/15/2019 03:59:38",
          "content": "<p>I havent directly measured but my first commit seems to take 3-5 minutes and my submit does seem to be about 10x that time. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 695887,
          "author_name": "jonasmatuzas",
          "author_url": "",
          "post_date": "12/15/2019 18:57:22",
          "content": "<p>my evaluation is also 10x</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "695323": "I made two submissions today with kernels that are correctly committed, and with enough room in terms of running time and RAM usage. Specifically, I made sure to save generated images in a temporary folder (/kaggle/tmp, so it's not in the output), so that i can run predictions directly by loading mini batches from disk.\n\nHowever, I ran into \"Notebook Exceeded Allowed Compute\" errors in both cases. Is that likely to be a RAM usage problem, a disk usage problem, or anything else relating to the size of the input?\n\nWould love to have your thoughts if you ran into this problem or solved it for this competition!",
    "695338": "Have not run into compute problems but be careful because the public test set which runs when we submit is much larger than the public validation set we have been given",
    "695385": "What's the size ratio? Should be less at most 10x no?",
    "695393": "I havent directly measured but my first commit seems to take 3-5 minutes and my submit does seem to be about 10x that time.",
    "695887": "my evaluation is also 10x"
  },
  "source": "meta"
}