{
  "id": 386611,
  "title": "Question on submission",
  "url": "/competitions/nfl-player-contact-detection/discussion/386611",
  "author_name": "",
  "post_date": "2023-02-13T21:34:54.936889500Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Okay, this would be my first submission, so please bear my naivety.<br>\nI am not sure how final submission would work, do I attach my trained model's checkpoint with it, or simply the code? And what does sample submission.csv imply? From what I know the model will be tested on a new data points, is model supposed to output in the format of sample submission? please help me out here, TIA.</p>",
  "messages": [
    {
      "id": "2142895",
      "postDate": "02/13/2023 21:34:54",
      "content": "<p>Okay, this would be my first submission, so please bear my naivety.<br>\nI am not sure how final submission would work, do I attach my trained model's checkpoint with it, or simply the code? And what does sample submission.csv imply? From what I know the model will be tested on a new data points, is model supposed to output in the format of sample submission? please help me out here, TIA.</p>",
      "rawMarkdown": "Okay, this would be my first submission, so please bear my naivety.\nI am not sure how final submission would work, do I attach my trained model's checkpoint with it, or simply the code? And what does sample submission.csv imply? From what I know the model will be tested on a new data points, is model supposed to output in the format of sample submission? please help me out here, TIA.",
      "votes": null
    },
    {
      "id": "2143016",
      "postDate": "02/14/2023 01:02:47",
      "content": "<p>You need to create a notebook that reads from the test csv files and the test directory, does the inference, and then creates a submission.csv file with your predictions.</p>\n<p>I find the best way to get familiar with the process is to see a working notebook.  Check out the \"code\" tab for several submission notebooks, including this one: <a href=\"https://www.kaggle.com/code/royalacecat/nfl-2-5d-cnn\" target=\"_blank\">https://www.kaggle.com/code/royalacecat/nfl-2-5d-cnn</a> (the current top scoring public notebook).</p>\n<p>While you're creating the notebook, you'll be able to see 2 plays in the test folder. Then you save it and it re-runs from top to bottom on those 2 plays to make sure there's no errors, and once that's complete, <em>then</em> you can submit it as a prediction, where it will run on all 61 unseen plays.</p>",
      "rawMarkdown": "You need to create a notebook that reads from the test csv files and the test directory, does the inference, and then creates a submission.csv file with your predictions.\n\nI find the best way to get familiar with the process is to see a working notebook.  Check out the \"code\" tab for several submission notebooks, including this one: https://www.kaggle.com/code/royalacecat/nfl-2-5d-cnn (the current top scoring public notebook).\n\nWhile you're creating the notebook, you'll be able to see 2 plays in the test folder. Then you save it and it re-runs from top to bottom on those 2 plays to make sure there's no errors, and once that's complete, _then_ you can submit it as a prediction, where it will run on all 61 unseen plays.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2143016,
      "author_name": "chris62",
      "author_url": "",
      "post_date": "02/14/2023 01:02:47",
      "content": "<p>You need to create a notebook that reads from the test csv files and the test directory, does the inference, and then creates a submission.csv file with your predictions.</p>\n<p>I find the best way to get familiar with the process is to see a working notebook.  Check out the \"code\" tab for several submission notebooks, including this one: <a href=\"https://www.kaggle.com/code/royalacecat/nfl-2-5d-cnn\" target=\"_blank\">https://www.kaggle.com/code/royalacecat/nfl-2-5d-cnn</a> (the current top scoring public notebook).</p>\n<p>While you're creating the notebook, you'll be able to see 2 plays in the test folder. Then you save it and it re-runs from top to bottom on those 2 plays to make sure there's no errors, and once that's complete, <em>then</em> you can submit it as a prediction, where it will run on all 61 unseen plays.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2142895": "Okay, this would be my first submission, so please bear my naivety.\nI am not sure how final submission would work, do I attach my trained model's checkpoint with it, or simply the code? And what does sample submission.csv imply? From what I know the model will be tested on a new data points, is model supposed to output in the format of sample submission? please help me out here, TIA.",
    "2143016": "You need to create a notebook that reads from the test csv files and the test directory, does the inference, and then creates a submission.csv file with your predictions.\n\nI find the best way to get familiar with the process is to see a working notebook.  Check out the \"code\" tab for several submission notebooks, including this one: https://www.kaggle.com/code/royalacecat/nfl-2-5d-cnn (the current top scoring public notebook).\n\nWhile you're creating the notebook, you'll be able to see 2 plays in the test folder. Then you save it and it re-runs from top to bottom on those 2 plays to make sure there's no errors, and once that's complete, _then_ you can submit it as a prediction, where it will run on all 61 unseen plays."
  },
  "source": "meta"
}