{
  "id": 667779,
  "title": "Asking if we can import notebook configured in jupyter notebook and submit",
  "url": "/competitions/deep-past-initiative-machine-translation/discussion/667779",
  "author_name": "",
  "post_date": "2026-01-14T07:28:35.075312600Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I have an A100 DGX spark on me , and I want to train models on it but confused about if i can then submit the prediction in the format </p>",
  "messages": [
    {
      "id": "3390943",
      "postDate": "01/14/2026 07:28:35",
      "content": "<p>I have an A100 DGX spark on me , and I want to train models on it but confused about if i can then submit the prediction in the format </p>",
      "rawMarkdown": "I have an A100 DGX spark on me , and I want to train models on it but confused about if i can then submit the prediction in the format",
      "votes": null
    },
    {
      "id": "3391059",
      "postDate": "01/14/2026 10:37:11",
      "content": "<p>I'm not sure I completely understood your question, hopefully this answers it: You should train the model wherever (your machine, other GPU providers, Kaggle) and then upload its weights or checkpoint to Kaggle. Then, you can create a separate notebook here which will only do the inference part. </p>\n<p>Once that script is ready, it should read in <code>test.csv</code> and produce <code>submission.csv</code> as output. After that works, you should click on \"Submit to competition\" button. </p>",
      "rawMarkdown": "I'm not sure I completely understood your question, hopefully this answers it: You should train the model wherever (your machine, other GPU providers, Kaggle) and then upload its weights or checkpoint to Kaggle. Then, you can create a separate notebook here which will only do the inference part. \n\nOnce that script is ready, it should read in `test.csv` and produce `submission.csv` as output. After that works, you should click on \"Submit to competition\" button.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3391059,
      "author_name": "acraevschi",
      "author_url": "",
      "post_date": "01/14/2026 10:37:11",
      "content": "<p>I'm not sure I completely understood your question, hopefully this answers it: You should train the model wherever (your machine, other GPU providers, Kaggle) and then upload its weights or checkpoint to Kaggle. Then, you can create a separate notebook here which will only do the inference part. </p>\n<p>Once that script is ready, it should read in <code>test.csv</code> and produce <code>submission.csv</code> as output. After that works, you should click on \"Submit to competition\" button. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3390943": "I have an A100 DGX spark on me , and I want to train models on it but confused about if i can then submit the prediction in the format",
    "3391059": "I'm not sure I completely understood your question, hopefully this answers it: You should train the model wherever (your machine, other GPU providers, Kaggle) and then upload its weights or checkpoint to Kaggle. Then, you can create a separate notebook here which will only do the inference part. \n\nOnce that script is ready, it should read in `test.csv` and produce `submission.csv` as output. After that works, you should click on \"Submit to competition\" button."
  },
  "source": "meta"
}