{
  "id": 572737,
  "title": "more compute than kaggle",
  "url": "/competitions/byu-locating-bacterial-flagellar-motors-2025/discussion/572737",
  "author_name": "",
  "post_date": "2025-04-11T05:10:17.932549300Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Are we allowed to train model outside kaggle and use it in submission? Are people using more compute, or the 12 hours kaggle model should be enough? </p>",
  "messages": [
    {
      "id": "3176211",
      "postDate": "04/11/2025 05:10:17",
      "content": "<p>Are we allowed to train model outside kaggle and use it in submission? Are people using more compute, or the 12 hours kaggle model should be enough? </p>",
      "rawMarkdown": "Are we allowed to train model outside kaggle and use it in submission? Are people using more compute, or the 12 hours kaggle model should be enough?",
      "votes": null
    },
    {
      "id": "3176228",
      "postDate": "04/11/2025 05:39:06",
      "content": "<p>Yes, we are allowed to train the model outside Kaggle. But inference will still be on the Kaggle's GPU and should take not longer than 12 hours.</p>",
      "rawMarkdown": "Yes, we are allowed to train the model outside Kaggle. But inference will still be on the Kaggle's GPU and should take not longer than 12 hours.",
      "votes": null
    },
    {
      "id": "3176239",
      "postDate": "04/11/2025 05:59:30",
      "content": "<p>Of course this is allowed and is a preferred option for most assignments  - one needs to use a PC with a big GPU to be able to experiment quickly and successfully. <br>\nIt is a good practice to train outside of Kaggle and infer here on Kaggle and submit (infer == make test set predictions only)<br>\n<a href=\"https://www.kaggle.com/thisarmin\" target=\"_blank\">@thisarmin</a> </p>",
      "rawMarkdown": "Of course this is allowed and is a preferred option for most assignments  - one needs to use a PC with a big GPU to be able to experiment quickly and successfully. \nIt is a good practice to train outside of Kaggle and infer here on Kaggle and submit (infer == make test set predictions only)\n@thisarmin",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3176228,
      "author_name": "eikyou",
      "author_url": "",
      "post_date": "04/11/2025 05:39:06",
      "content": "<p>Yes, we are allowed to train the model outside Kaggle. But inference will still be on the Kaggle's GPU and should take not longer than 12 hours.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3176239,
      "author_name": "ravi20076",
      "author_url": "",
      "post_date": "04/11/2025 05:59:30",
      "content": "<p>Of course this is allowed and is a preferred option for most assignments  - one needs to use a PC with a big GPU to be able to experiment quickly and successfully. <br>\nIt is a good practice to train outside of Kaggle and infer here on Kaggle and submit (infer == make test set predictions only)<br>\n<a href=\"https://www.kaggle.com/thisarmin\" target=\"_blank\">@thisarmin</a> </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3176211": "Are we allowed to train model outside kaggle and use it in submission? Are people using more compute, or the 12 hours kaggle model should be enough?",
    "3176228": "Yes, we are allowed to train the model outside Kaggle. But inference will still be on the Kaggle's GPU and should take not longer than 12 hours.",
    "3176239": "Of course this is allowed and is a preferred option for most assignments  - one needs to use a PC with a big GPU to be able to experiment quickly and successfully. \nIt is a good practice to train outside of Kaggle and infer here on Kaggle and submit (infer == make test set predictions only)\n@thisarmin"
  },
  "source": "meta"
}