{
  "id": 301434,
  "title": "Offline training and inference from saved models",
  "url": "/competitions/g-research-crypto-forecasting/discussion/301434",
  "author_name": "",
  "post_date": "2022-01-17T18:07:41.602735600Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>When save models are stored in input . If this is possible for this competition ? That was possible for Janestreet</p>",
  "messages": [
    {
      "id": "1653624",
      "postDate": "01/17/2022 18:07:41",
      "content": "<p>When save models are stored in input . If this is possible for this competition ? That was possible for Janestreet</p>",
      "rawMarkdown": "When save models are stored in input . If this is possible for this competition ? That was possible for Janestreet",
      "votes": null
    },
    {
      "id": "1653857",
      "postDate": "01/17/2022 23:38:08",
      "content": "<p>You can save the model parameters as a kaggle dataset and import the dataset into kaggle notebooks.  I think you can also use the output of a notebook as the input to another notebook.  So, you could run a notebook to create the file that stores the model params</p>",
      "rawMarkdown": "You can save the model parameters as a kaggle dataset and import the dataset into kaggle notebooks.  I think you can also use the output of a notebook as the input to another notebook.  So, you could run a notebook to create the file that stores the model params",
      "votes": null
    },
    {
      "id": "1654954",
      "postDate": "01/19/2022 01:42:32",
      "content": "<p>The rules of the competition state:</p>\n<blockquote>\n  <p>This is a Code Competition</p>\n  <p>CPU Notebook &lt;= 9 hours run-time</p>\n  <p>GPU Notebook &lt;= 9 hours run-time</p>\n  <p>Internet access disabled</p>\n  <p>Freely &amp; publicly available external data is allowed, including pre-trained models</p>\n  <p>Submission file must be named submission.csv</p>\n</blockquote>\n<p>Yes, this is allowed. As <a href=\"https://www.kaggle.com/petersk20\" target=\"_blank\">@petersk20</a> said, you can do it in multiple ways.</p>",
      "rawMarkdown": "The rules of the competition state:\n> This is a Code Competition\n\n> CPU Notebook <= 9 hours run-time\n\n> GPU Notebook <= 9 hours run-time\n\n> Internet access disabled\n\n> Freely & publicly available external data is allowed, including pre-trained models\n\n> Submission file must be named submission.csv\n\nYes, this is allowed. As @petersk20 said, you can do it in multiple ways.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1653857,
      "author_name": "petersk20",
      "author_url": "",
      "post_date": "01/17/2022 23:38:08",
      "content": "<p>You can save the model parameters as a kaggle dataset and import the dataset into kaggle notebooks.  I think you can also use the output of a notebook as the input to another notebook.  So, you could run a notebook to create the file that stores the model params</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1654954,
      "author_name": "",
      "author_url": "",
      "post_date": "01/19/2022 01:42:32",
      "content": "<p>The rules of the competition state:</p>\n<blockquote>\n  <p>This is a Code Competition</p>\n  <p>CPU Notebook &lt;= 9 hours run-time</p>\n  <p>GPU Notebook &lt;= 9 hours run-time</p>\n  <p>Internet access disabled</p>\n  <p>Freely &amp; publicly available external data is allowed, including pre-trained models</p>\n  <p>Submission file must be named submission.csv</p>\n</blockquote>\n<p>Yes, this is allowed. As <a href=\"https://www.kaggle.com/petersk20\" target=\"_blank\">@petersk20</a> said, you can do it in multiple ways.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1653624": "When save models are stored in input . If this is possible for this competition ? That was possible for Janestreet",
    "1653857": "You can save the model parameters as a kaggle dataset and import the dataset into kaggle notebooks.  I think you can also use the output of a notebook as the input to another notebook.  So, you could run a notebook to create the file that stores the model params",
    "1654954": "The rules of the competition state:\n> This is a Code Competition\n\n> CPU Notebook <= 9 hours run-time\n\n> GPU Notebook <= 9 hours run-time\n\n> Internet access disabled\n\n> Freely & publicly available external data is allowed, including pre-trained models\n\n> Submission file must be named submission.csv\n\nYes, this is allowed. As @petersk20 said, you can do it in multiple ways."
  },
  "source": "meta"
}