{
  "id": 457693,
  "title": "Do models have to be trained within the Notebook execution?",
  "url": "/competitions/blood-vessel-segmentation/discussion/457693",
  "author_name": "",
  "post_date": "2023-11-26T08:42:21.104930500Z",
  "votes": 4,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi,</p>\n<p>For competitions that require a notebook submission, are we allowed to train a model externally and upload it to the notebook before submitting, or does training have to be performed within the notebook?</p>\n<p>It's not clear to me whether the notebook limitations are meant for both the training and testing phase, or just for the testing phase.</p>\n<p>Thanks,<br>\nShayan</p>",
  "messages": [
    {
      "id": "2538540",
      "postDate": "11/26/2023 08:42:21",
      "content": "<p>Hi,</p>\n<p>For competitions that require a notebook submission, are we allowed to train a model externally and upload it to the notebook before submitting, or does training have to be performed within the notebook?</p>\n<p>It's not clear to me whether the notebook limitations are meant for both the training and testing phase, or just for the testing phase.</p>\n<p>Thanks,<br>\nShayan</p>",
      "rawMarkdown": "Hi,\n\nFor competitions that require a notebook submission, are we allowed to train a model externally and upload it to the notebook before submitting, or does training have to be performed within the notebook?\n\nIt's not clear to me whether the notebook limitations are meant for both the training and testing phase, or just for the testing phase.\n\nThanks,\nShayan",
      "votes": null
    },
    {
      "id": "2538550",
      "postDate": "11/26/2023 08:56:36",
      "content": "<p>You are free to train your model outside or inside the notebook. Most people will indeed only use the submission notebook as an inference notebook.</p>",
      "rawMarkdown": "You are free to train your model outside or inside the notebook. Most people will indeed only use the submission notebook as an inference notebook.",
      "votes": null
    },
    {
      "id": "2538560",
      "postDate": "11/26/2023 09:09:49",
      "content": "<p>Your models may be trained outside of Kaggle preferably and inferred in a Kaggle kernel, using the trained model/ models as a dataset input. This will ensure your submission is clean and you do not overshoot the time constraints associated with the evaluation process <a href=\"https://www.kaggle.com/moejoe\" target=\"_blank\">@moejoe</a> </p>",
      "rawMarkdown": "Your models may be trained outside of Kaggle preferably and inferred in a Kaggle kernel, using the trained model/ models as a dataset input. This will ensure your submission is clean and you do not overshoot the time constraints associated with the evaluation process @moejoe",
      "votes": null
    },
    {
      "id": "2547979",
      "postDate": "12/04/2023 04:19:57",
      "content": "<p>You can train the model everywhere but have to use kaggle notebook for inference</p>",
      "rawMarkdown": "You can train the model everywhere but have to use kaggle notebook for inference",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2538550,
      "author_name": "optimo",
      "author_url": "",
      "post_date": "11/26/2023 08:56:36",
      "content": "<p>You are free to train your model outside or inside the notebook. Most people will indeed only use the submission notebook as an inference notebook.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2538560,
      "author_name": "ravi20076",
      "author_url": "",
      "post_date": "11/26/2023 09:09:49",
      "content": "<p>Your models may be trained outside of Kaggle preferably and inferred in a Kaggle kernel, using the trained model/ models as a dataset input. This will ensure your submission is clean and you do not overshoot the time constraints associated with the evaluation process <a href=\"https://www.kaggle.com/moejoe\" target=\"_blank\">@moejoe</a> </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2547979,
      "author_name": "trytobebetter",
      "author_url": "",
      "post_date": "12/04/2023 04:19:57",
      "content": "<p>You can train the model everywhere but have to use kaggle notebook for inference</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2538540": "Hi,\n\nFor competitions that require a notebook submission, are we allowed to train a model externally and upload it to the notebook before submitting, or does training have to be performed within the notebook?\n\nIt's not clear to me whether the notebook limitations are meant for both the training and testing phase, or just for the testing phase.\n\nThanks,\nShayan",
    "2538550": "You are free to train your model outside or inside the notebook. Most people will indeed only use the submission notebook as an inference notebook.",
    "2538560": "Your models may be trained outside of Kaggle preferably and inferred in a Kaggle kernel, using the trained model/ models as a dataset input. This will ensure your submission is clean and you do not overshoot the time constraints associated with the evaluation process @moejoe",
    "2547979": "You can train the model everywhere but have to use kaggle notebook for inference"
  },
  "source": "meta"
}