{
  "id": 491555,
  "title": "Reminder - Save some GPU Quota for last submissions",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/491555",
  "author_name": "something4kag",
  "post_date": "2024-04-06T09:37:35.763000",
  "votes": 11,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Just a reminder that some GPU Quota is needed to do any last submissions in case not aware of this. So if you are confident that your inference code is working fine - To avoid running everything twice you can check this environment variable -</p>\n<blockquote>\n  <p>if os.getenv('KAGGLE_IS_COMPETITION_RERUN'):</p>\n</blockquote>\n<p>If it is not a rerun so False, then you can just copy the sample submission to /kaggle/working or read it and write it there so you have an output file to submit and bypass code to save GPU.  If it is a rerun so True (as it is in Private when submitted) it will do your test inference with the model and whatever else is needed in your code. </p>\n<p>Also, just an FYI: </p>\n<ul>\n<li>Notebooks for submissions do need to be Save and Commit (not Quick Save) </li>\n<li>To select for final LB they have to finish running before the end of competition Final Submission Deadline.</li>\n</ul>\n<p>Good Luck to everyone!</p>",
  "messages": [
    {
      "id": 2738355,
      "postDate": "2024-04-06T09:37:35.763Z",
      "content": "<p>Just a reminder that some GPU Quota is needed to do any last submissions in case not aware of this. So if you are confident that your inference code is working fine - To avoid running everything twice you can check this environment variable -</p>\n<blockquote>\n  <p>if os.getenv('KAGGLE_IS_COMPETITION_RERUN'):</p>\n</blockquote>\n<p>If it is not a rerun so False, then you can just copy the sample submission to /kaggle/working or read it and write it there so you have an output file to submit and bypass code to save GPU.  If it is a rerun so True (as it is in Private when submitted) it will do your test inference with the model and whatever else is needed in your code. </p>\n<p>Also, just an FYI: </p>\n<ul>\n<li>Notebooks for submissions do need to be Save and Commit (not Quick Save) </li>\n<li>To select for final LB they have to finish running before the end of competition Final Submission Deadline.</li>\n</ul>\n<p>Good Luck to everyone!</p>",
      "rawMarkdown": "Just a reminder that some GPU Quota is needed to do any last submissions in case not aware of this. So if you are confident that your inference code is working fine - To avoid running everything twice you can check this environment variable -\n\n>if os.getenv('KAGGLE_IS_COMPETITION_RERUN'):\n\nIf it is not a rerun so False, then you can just copy the sample submission to /kaggle/working or read it and write it there so you have an output file to submit and bypass code to save GPU.  If it is a rerun so True (as it is in Private when submitted) it will do your test inference with the model and whatever else is needed in your code. \n\nAlso, just an FYI: \n- Notebooks for submissions do need to be Save and Commit (not Quick Save) \n- To select for final LB they have to finish running before the end of competition Final Submission Deadline.\n\nGood Luck to everyone!",
      "votes": 11
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2738355": "Just a reminder that some GPU Quota is needed to do any last submissions in case not aware of this. So if you are confident that your inference code is working fine - To avoid running everything twice you can check this environment variable -\n\n>if os.getenv('KAGGLE_IS_COMPETITION_RERUN'):\n\nIf it is not a rerun so False, then you can just copy the sample submission to /kaggle/working or read it and write it there so you have an output file to submit and bypass code to save GPU.  If it is a rerun so True (as it is in Private when submitted) it will do your test inference with the model and whatever else is needed in your code. \n\nAlso, just an FYI: \n- Notebooks for submissions do need to be Save and Commit (not Quick Save) \n- To select for final LB they have to finish running before the end of competition Final Submission Deadline.\n\nGood Luck to everyone!"
  }
}