{
  "id": 217689,
  "title": "GPU not being assigned before restart",
  "url": "/competitions/rfcx-species-audio-detection/discussion/217689",
  "author_name": "ryches",
  "post_date": "2021-02-07T23:04:01.778000",
  "votes": 2,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Occasionally I will get the warning message from kaggle saying GPU was not used on previous run so it was removed from the notebook. If I click the option to keep it enabled then the notebook will restart and show that it has GPU, but it wont actually be there. It shows it as having 0 memory capacity and when I try to do any gpu operation it fails or just goes very slow because it falls back on CPU. </p>\n<p><img src=\"https://i.imgur.com/oTswRXN.png\" alt=\"\"></p>\n<p>A simple restart of the notebook then fixes this, but I figured I would at least report the bug. Sometimes it can eat considerable amount of time if there is some data prep step and I only find that the GPU wasn't actually available by the end of the notebook when training begins. </p>\n<p><img src=\"https://i.imgur.com/KXbGzOL.png\" alt=\"\"></p>",
  "messages": [
    {
      "id": 1190696,
      "postDate": "2021-02-07T23:04:01.780Z",
      "content": "<p>Occasionally I will get the warning message from kaggle saying GPU was not used on previous run so it was removed from the notebook. If I click the option to keep it enabled then the notebook will restart and show that it has GPU, but it wont actually be there. It shows it as having 0 memory capacity and when I try to do any gpu operation it fails or just goes very slow because it falls back on CPU. </p>\n<p><img src=\"https://i.imgur.com/oTswRXN.png\" alt=\"\"></p>\n<p>A simple restart of the notebook then fixes this, but I figured I would at least report the bug. Sometimes it can eat considerable amount of time if there is some data prep step and I only find that the GPU wasn't actually available by the end of the notebook when training begins. </p>\n<p><img src=\"https://i.imgur.com/KXbGzOL.png\" alt=\"\"></p>",
      "rawMarkdown": "Occasionally I will get the warning message from kaggle saying GPU was not used on previous run so it was removed from the notebook. If I click the option to keep it enabled then the notebook will restart and show that it has GPU, but it wont actually be there. It shows it as having 0 memory capacity and when I try to do any gpu operation it fails or just goes very slow because it falls back on CPU. \n\n![](https://i.imgur.com/oTswRXN.png)\n\n\nA simple restart of the notebook then fixes this, but I figured I would at least report the bug. Sometimes it can eat considerable amount of time if there is some data prep step and I only find that the GPU wasn't actually available by the end of the notebook when training begins. \n\n![](https://i.imgur.com/KXbGzOL.png)",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1190696": "Occasionally I will get the warning message from kaggle saying GPU was not used on previous run so it was removed from the notebook. If I click the option to keep it enabled then the notebook will restart and show that it has GPU, but it wont actually be there. It shows it as having 0 memory capacity and when I try to do any gpu operation it fails or just goes very slow because it falls back on CPU. \n\n![](https://i.imgur.com/oTswRXN.png)\n\n\nA simple restart of the notebook then fixes this, but I figured I would at least report the bug. Sometimes it can eat considerable amount of time if there is some data prep step and I only find that the GPU wasn't actually available by the end of the notebook when training begins. \n\n![](https://i.imgur.com/KXbGzOL.png)"
  }
}