{
  "id": 405628,
  "title": "May someone explain code submissions?",
  "url": "/competitions/asl-signs/discussion/405628",
  "author_name": "",
  "post_date": "2023-04-28T16:03:14.906225900Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I am currently finishing up my final submission. </p>\n<p>However, I was wondering whether I can save my models into a public dataset and then submit a notebook which loads and submits them. The dataset would not get a lot of attention and therefore nobody would copy my solution. I would avoid long training in my submission, etc. </p>\n<p>May someone tell us whether this is possible and what other thighs should be considered in code submissions? </p>",
  "messages": [
    {
      "id": "2238555",
      "postDate": "04/28/2023 16:03:14",
      "content": "<p>I am currently finishing up my final submission. </p>\n<p>However, I was wondering whether I can save my models into a public dataset and then submit a notebook which loads and submits them. The dataset would not get a lot of attention and therefore nobody would copy my solution. I would avoid long training in my submission, etc. </p>\n<p>May someone tell us whether this is possible and what other thighs should be considered in code submissions? </p>",
      "rawMarkdown": "I am currently finishing up my final submission. \n\nHowever, I was wondering whether I can save my models into a public dataset and then submit a notebook which loads and submits them. The dataset would not get a lot of attention and therefore nobody would copy my solution. I would avoid long training in my submission, etc. \n\nMay someone tell us whether this is possible and what other thighs should be considered in code submissions?",
      "votes": null
    },
    {
      "id": "2238607",
      "postDate": "04/28/2023 17:03:26",
      "content": "<p>Why don't use a private dataset? I think publishing a high-score final model would be a good idea. Personally, I upload tflite model to a private dataset and then submit it via private notebook</p>",
      "rawMarkdown": "Why don't use a private dataset? I think publishing a high-score final model would be a good idea. Personally, I upload tflite model to a private dataset and then submit it via private notebook",
      "votes": null
    },
    {
      "id": "2238619",
      "postDate": "04/28/2023 17:10:09",
      "content": "<p>because it says in the \"code requirements\" that you may use public available sources… I mean, I wouldn't like to submit a notebook which uses my private dataset and then be disqualificated because the notebook does not use only open sources.  </p>",
      "rawMarkdown": "because it says in the \"code requirements\" that you may use public available sources... I mean, I wouldn't like to submit a notebook which uses my private dataset and then be disqualificated because the notebook does not use only open sources.",
      "votes": null
    },
    {
      "id": "2238733",
      "postDate": "04/28/2023 18:55:57",
      "content": "<p><a href=\"https://www.kaggle.com/gzguevara\" target=\"_blank\">@gzguevara</a> you can upload your locally trained model as a private dataset. It is the way to do it. You just can't train it on data that is not allowed.</p>",
      "rawMarkdown": "gzguevara you can upload your locally trained model as a private dataset. It is the way to do it. You just can't train it on data that is not allowed.",
      "votes": null
    },
    {
      "id": "2238802",
      "postDate": "04/28/2023 20:04:22",
      "content": "<p>You are not allowed to use a private dataset as a private data source for your model training. But there are no limits to using \"the kaggle private dataset\" feature. Feel free to upload your model in such a way. This is the best way to go</p>",
      "rawMarkdown": "You are not allowed to use a private dataset as a private data source for your model training. But there are no limits to using \"the kaggle private dataset\" feature. Feel free to upload your model in such a way. This is the best way to go",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2238607,
      "author_name": "meowmeowmeowmeowmeow",
      "author_url": "",
      "post_date": "04/28/2023 17:03:26",
      "content": "<p>Why don't use a private dataset? I think publishing a high-score final model would be a good idea. Personally, I upload tflite model to a private dataset and then submit it via private notebook</p>",
      "votes": null,
      "replies": [
        {
          "id": 2238619,
          "author_name": "gzguevara",
          "author_url": "",
          "post_date": "04/28/2023 17:10:09",
          "content": "<p>because it says in the \"code requirements\" that you may use public available sources… I mean, I wouldn't like to submit a notebook which uses my private dataset and then be disqualificated because the notebook does not use only open sources.  </p>",
          "votes": null,
          "replies": [
            {
              "id": 2238733,
              "author_name": "crodoc",
              "author_url": "",
              "post_date": "04/28/2023 18:55:57",
              "content": "<p><a href=\"https://www.kaggle.com/gzguevara\" target=\"_blank\">@gzguevara</a> you can upload your locally trained model as a private dataset. It is the way to do it. You just can't train it on data that is not allowed.</p>",
              "votes": null,
              "replies": []
            },
            {
              "id": 2238802,
              "author_name": "meowmeowmeowmeowmeow",
              "author_url": "",
              "post_date": "04/28/2023 20:04:22",
              "content": "<p>You are not allowed to use a private dataset as a private data source for your model training. But there are no limits to using \"the kaggle private dataset\" feature. Feel free to upload your model in such a way. This is the best way to go</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2238555": "I am currently finishing up my final submission. \n\nHowever, I was wondering whether I can save my models into a public dataset and then submit a notebook which loads and submits them. The dataset would not get a lot of attention and therefore nobody would copy my solution. I would avoid long training in my submission, etc. \n\nMay someone tell us whether this is possible and what other thighs should be considered in code submissions?",
    "2238607": "Why don't use a private dataset? I think publishing a high-score final model would be a good idea. Personally, I upload tflite model to a private dataset and then submit it via private notebook",
    "2238619": "because it says in the \"code requirements\" that you may use public available sources... I mean, I wouldn't like to submit a notebook which uses my private dataset and then be disqualificated because the notebook does not use only open sources.",
    "2238733": "gzguevara you can upload your locally trained model as a private dataset. It is the way to do it. You just can't train it on data that is not allowed.",
    "2238802": "You are not allowed to use a private dataset as a private data source for your model training. But there are no limits to using \"the kaggle private dataset\" feature. Feel free to upload your model in such a way. This is the best way to go"
  },
  "source": "meta"
}