{
  "id": 255077,
  "title": "Any Help on Notebook Threw Exception",
  "url": "/competitions/siim-covid19-detection/discussion/255077",
  "author_name": "",
  "post_date": "2021-07-25T13:59:49.749523100Z",
  "votes": null,
  "comment_count": 5,
  "views": 0,
  "content": "<p>I get above error during inference  close to the time of its completion.<br>\nAny help highly appreciated <br>\nThanks in advance. </p>",
  "messages": [
    {
      "id": "1399606",
      "postDate": "07/25/2021 13:59:49",
      "content": "<p>I get above error during inference  close to the time of its completion.<br>\nAny help highly appreciated <br>\nThanks in advance. </p>",
      "rawMarkdown": "I get above error during inference  close to the time of its completion.\nAny help highly appreciated \nThanks in advance.",
      "votes": null
    },
    {
      "id": "1400079",
      "postDate": "07/26/2021 02:46:26",
      "content": "<p>May be because the data are not the same between you run the notebook and submit the notebook.</p>",
      "rawMarkdown": "May be because the data are not the same between you run the notebook and submit the notebook.",
      "votes": null
    },
    {
      "id": "1400765",
      "postDate": "07/26/2021 15:05:34",
      "content": "<p>It is some error in your code that you have to find.</p>\n<p>In my case, I found that I had some parameters in the code which worked fine on the sample submission set but did not generalize to the public leaderboard submission. For example if you are looping through some dataset, you need to dynamically determine the number of rows rather than looping through a constant value</p>\n<p><code>for i in range(1200):</code> will throw an error on submission</p>\n<p><code>for i in range(len(df)):</code> will work better in this example</p>\n<p>Hope that helps in some way!</p>",
      "rawMarkdown": "It is some error in your code that you have to find.\n\nIn my case, I found that I had some parameters in the code which worked fine on the sample submission set but did not generalize to the public leaderboard submission. For example if you are looping through some dataset, you need to dynamically determine the number of rows rather than looping through a constant value\n\n`for i in range(1200):` will throw an error on submission\n\n`for i in range(len(df)):` will work better in this example\n\nHope that helps in some way!",
      "votes": null
    },
    {
      "id": "1400860",
      "postDate": "07/26/2021 16:51:38",
      "content": "<p>Hi, there are too many models and memory cannot hold them all. It may look that the notebook can be saved without error, but during inference on the hidden dataset which has a larger data size, more memory will be required during running. I know this because I once met the same thing. After removing one model, my notebook submission did not throw exception. Still, it is about memory.</p>",
      "rawMarkdown": "Hi, there are too many models and memory cannot hold them all. It may look that the notebook can be saved without error, but during inference on the hidden dataset which has a larger data size, more memory will be required during running. I know this because I once met the same thing. After removing one model, my notebook submission did not throw exception. Still, it is about memory.",
      "votes": null
    },
    {
      "id": "1407631",
      "postDate": "08/02/2021 04:11:35",
      "content": "<p>Is there anyway to check which part of my code is causing this error? This is really frustrating </p>",
      "rawMarkdown": "Is there anyway to check which part of my code is causing this error? This is really frustrating",
      "votes": null
    },
    {
      "id": "1407650",
      "postDate": "08/02/2021 04:32:12",
      "content": "<p>In my case , I made some mistakes(overflow) when normalizing the data,<br>\nI predicted all public test data without error , but failed in submitting.  </p>",
      "rawMarkdown": "In my case , I made some mistakes(overflow) when normalizing the data,\nI predicted all public test data without error , but failed in submitting.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1400079,
      "author_name": "wyyyywwy",
      "author_url": "",
      "post_date": "07/26/2021 02:46:26",
      "content": "<p>May be because the data are not the same between you run the notebook and submit the notebook.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1400765,
      "author_name": "illidan7",
      "author_url": "",
      "post_date": "07/26/2021 15:05:34",
      "content": "<p>It is some error in your code that you have to find.</p>\n<p>In my case, I found that I had some parameters in the code which worked fine on the sample submission set but did not generalize to the public leaderboard submission. For example if you are looping through some dataset, you need to dynamically determine the number of rows rather than looping through a constant value</p>\n<p><code>for i in range(1200):</code> will throw an error on submission</p>\n<p><code>for i in range(len(df)):</code> will work better in this example</p>\n<p>Hope that helps in some way!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1400860,
      "author_name": "yus002",
      "author_url": "",
      "post_date": "07/26/2021 16:51:38",
      "content": "<p>Hi, there are too many models and memory cannot hold them all. It may look that the notebook can be saved without error, but during inference on the hidden dataset which has a larger data size, more memory will be required during running. I know this because I once met the same thing. After removing one model, my notebook submission did not throw exception. Still, it is about memory.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1407631,
      "author_name": "sunniem",
      "author_url": "",
      "post_date": "08/02/2021 04:11:35",
      "content": "<p>Is there anyway to check which part of my code is causing this error? This is really frustrating </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1407650,
      "author_name": "atom1231",
      "author_url": "",
      "post_date": "08/02/2021 04:32:12",
      "content": "<p>In my case , I made some mistakes(overflow) when normalizing the data,<br>\nI predicted all public test data without error , but failed in submitting.  </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1399606": "I get above error during inference  close to the time of its completion.\nAny help highly appreciated \nThanks in advance.",
    "1400079": "May be because the data are not the same between you run the notebook and submit the notebook.",
    "1400765": "It is some error in your code that you have to find.\n\nIn my case, I found that I had some parameters in the code which worked fine on the sample submission set but did not generalize to the public leaderboard submission. For example if you are looping through some dataset, you need to dynamically determine the number of rows rather than looping through a constant value\n\n`for i in range(1200):` will throw an error on submission\n\n`for i in range(len(df)):` will work better in this example\n\nHope that helps in some way!",
    "1400860": "Hi, there are too many models and memory cannot hold them all. It may look that the notebook can be saved without error, but during inference on the hidden dataset which has a larger data size, more memory will be required during running. I know this because I once met the same thing. After removing one model, my notebook submission did not throw exception. Still, it is about memory.",
    "1407631": "Is there anyway to check which part of my code is causing this error? This is really frustrating",
    "1407650": "In my case , I made some mistakes(overflow) when normalizing the data,\nI predicted all public test data without error , but failed in submitting."
  },
  "source": "meta"
}