{
  "id": 679077,
  "title": "Clarification of Rules",
  "url": "/competitions/deep-past-initiative-machine-translation/discussion/679077",
  "author_name": "",
  "post_date": "2026-02-26T22:31:47.646086400Z",
  "votes": -1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I just want to make sure I understand the rules: Are we allowed to fine-tune a model outside of the submission notebook and then, say, inference off of it in the submission notebook? Similarly, can we transform the data in some way outside of the notebook and then utilize that data in the submission? Or can we only do these things if we make the data/fine-tuned model publicly available? </p>",
  "messages": [
    {
      "id": "3414489",
      "postDate": "02/26/2026 22:31:47",
      "content": "<p>I just want to make sure I understand the rules: Are we allowed to fine-tune a model outside of the submission notebook and then, say, inference off of it in the submission notebook? Similarly, can we transform the data in some way outside of the notebook and then utilize that data in the submission? Or can we only do these things if we make the data/fine-tuned model publicly available? </p>",
      "rawMarkdown": "I just want to make sure I understand the rules: Are we allowed to fine-tune a model outside of the submission notebook and then, say, inference off of it in the submission notebook? Similarly, can we transform the data in some way outside of the notebook and then utilize that data in the submission? Or can we only do these things if we make the data/fine-tuned model publicly available?",
      "votes": null
    },
    {
      "id": "3414492",
      "postDate": "02/26/2026 23:18:00",
      "content": "<p>You can train/fine-tune your model outside Kaggle. You just need to make sure your inference runs successfully in the submission notebook without internet access. The same applies to external data processing: you can preprocess data offline and use it in the notebook as long as you don’t connect to the internet during submission.good luck</p>",
      "rawMarkdown": "You can train/fine-tune your model outside Kaggle. You just need to make sure your inference runs successfully in the submission notebook without internet access. The same applies to external data processing: you can preprocess data offline and use it in the notebook as long as you don’t connect to the internet during submission.good luck",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3414492,
      "author_name": "qifeihhh666",
      "author_url": "",
      "post_date": "02/26/2026 23:18:00",
      "content": "<p>You can train/fine-tune your model outside Kaggle. You just need to make sure your inference runs successfully in the submission notebook without internet access. The same applies to external data processing: you can preprocess data offline and use it in the notebook as long as you don’t connect to the internet during submission.good luck</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3414489": "I just want to make sure I understand the rules: Are we allowed to fine-tune a model outside of the submission notebook and then, say, inference off of it in the submission notebook? Similarly, can we transform the data in some way outside of the notebook and then utilize that data in the submission? Or can we only do these things if we make the data/fine-tuned model publicly available?",
    "3414492": "You can train/fine-tune your model outside Kaggle. You just need to make sure your inference runs successfully in the submission notebook without internet access. The same applies to external data processing: you can preprocess data offline and use it in the notebook as long as you don’t connect to the internet during submission.good luck"
  },
  "source": "meta"
}