{
  "id": 443256,
  "title": "Notebook Out of Memory error on Submission ",
  "url": "/competitions/rsna-2023-abdominal-trauma-detection/discussion/443256",
  "author_name": "Prateek Sanghi Onward",
  "post_date": "2023-09-26T08:43:43.803000",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I'm getting Notebook Out of Memory error on Submission </p>\n<p>It states \"Your notebook requested more memory (RAM) than is available.\" </p>\n<p>if anyone also faced a similar issue and has fixed it please let me know</p>",
  "messages": [
    {
      "id": 2457610,
      "postDate": "2023-09-27T04:40:38.607Z",
      "content": "<p>I encountered such an issue.<br>\nFirst, I debugged by running the notebook on a train subset. <br>\nI found that the loading and pre-processing of the largest dicom volumes can consume a lot of memory and depending on the implementation may take more than the memory limit of 13 GB.<br>\nIf that is the root cause, one or more of following might help: window, resize and convert 2D dicom slices to uint8 as soon as loaded. use del to delete unneeded objects and gc.collect() to free memory asap.</p>",
      "rawMarkdown": "I encountered such an issue.\nFirst, I debugged by running the notebook on a train subset. \nI found that the loading and pre-processing of the largest dicom volumes can consume a lot of memory and depending on the implementation may take more than the memory limit of 13 GB.\nIf that is the root cause, one or more of following might help: window, resize and convert 2D dicom slices to uint8 as soon as loaded. use del to delete unneeded objects and gc.collect() to free memory asap.",
      "votes": 1,
      "replies": [
        {
          "id": 2459546,
          "postDate": "2023-09-28T09:02:18.223Z",
          "content": "<p>ok will give it a try </p>",
          "rawMarkdown": "ok will give it a try "
        }
      ]
    },
    {
      "id": 2456503,
      "postDate": "2023-09-26T08:43:43.803Z",
      "content": "<p>I'm getting Notebook Out of Memory error on Submission </p>\n<p>It states \"Your notebook requested more memory (RAM) than is available.\" </p>\n<p>if anyone also faced a similar issue and has fixed it please let me know</p>",
      "rawMarkdown": "I'm getting Notebook Out of Memory error on Submission \n\nIt states \"Your notebook requested more memory (RAM) than is available.\" \n\nif anyone also faced a similar issue and has fixed it please let me know",
      "votes": 1
    },
    {
      "id": 2456543,
      "postDate": "2023-09-26T09:08:28.653Z",
      "content": "<p>Are you using GPU?</p>",
      "rawMarkdown": "Are you using GPU?",
      "replies": [
        {
          "id": 2456559,
          "postDate": "2023-09-26T09:19:20.420Z",
          "content": "<p>yes im using the GPU </p>",
          "rawMarkdown": "yes im using the GPU "
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 2457610,
      "author_name": "WalkingMoose",
      "author_url": "",
      "post_date": "2023-09-27T04:40:38.607000",
      "content": "<p>I encountered such an issue.<br>\nFirst, I debugged by running the notebook on a train subset. <br>\nI found that the loading and pre-processing of the largest dicom volumes can consume a lot of memory and depending on the implementation may take more than the memory limit of 13 GB.<br>\nIf that is the root cause, one or more of following might help: window, resize and convert 2D dicom slices to uint8 as soon as loaded. use del to delete unneeded objects and gc.collect() to free memory asap.</p>",
      "votes": 1,
      "replies": [
        {
          "id": 2459546,
          "author_name": "Prateek Sanghi Onward",
          "author_url": "",
          "post_date": "2023-09-28T09:02:18.223000",
          "content": "<p>ok will give it a try </p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 2456543,
      "author_name": "Faysal Miah",
      "author_url": "",
      "post_date": "2023-09-26T09:08:28.653000",
      "content": "<p>Are you using GPU?</p>",
      "votes": 0,
      "replies": [
        {
          "id": 2456559,
          "author_name": "Prateek Sanghi Onward",
          "author_url": "",
          "post_date": "2023-09-26T09:19:20.420000",
          "content": "<p>yes im using the GPU </p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2457610": "I encountered such an issue.\nFirst, I debugged by running the notebook on a train subset. \nI found that the loading and pre-processing of the largest dicom volumes can consume a lot of memory and depending on the implementation may take more than the memory limit of 13 GB.\nIf that is the root cause, one or more of following might help: window, resize and convert 2D dicom slices to uint8 as soon as loaded. use del to delete unneeded objects and gc.collect() to free memory asap.",
    "2456503": "I'm getting Notebook Out of Memory error on Submission \n\nIt states \"Your notebook requested more memory (RAM) than is available.\" \n\nif anyone also faced a similar issue and has fixed it please let me know",
    "2456543": "Are you using GPU?"
  }
}