{
  "id": 394883,
  "title": "Query regarding training system",
  "url": "/competitions/asl-signs/discussion/394883",
  "author_name": "",
  "post_date": "2023-03-15T04:30:25.586718500Z",
  "votes": 3,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi, I'm new to kaggle competitions so I'm not sure if this would be allowed. Can I train a model locally on my computer, and upload the trained model as a zipped file via the Kaggle dataset upload option on the notebook. I've read this as a hack to upload our models to the kaggle notebook, but I'm not sure if that is allowed in competitions. Would appreciate some help on this!</p>",
  "messages": [
    {
      "id": "2182358",
      "postDate": "03/15/2023 04:30:25",
      "content": "<p>Hi, I'm new to kaggle competitions so I'm not sure if this would be allowed. Can I train a model locally on my computer, and upload the trained model as a zipped file via the Kaggle dataset upload option on the notebook. I've read this as a hack to upload our models to the kaggle notebook, but I'm not sure if that is allowed in competitions. Would appreciate some help on this!</p>",
      "rawMarkdown": "Hi, I'm new to kaggle competitions so I'm not sure if this would be allowed. Can I train a model locally on my computer, and upload the trained model as a zipped file via the Kaggle dataset upload option on the notebook. I've read this as a hack to upload our models to the kaggle notebook, but I'm not sure if that is allowed in competitions. Would appreciate some help on this!",
      "votes": null
    },
    {
      "id": "2182477",
      "postDate": "03/15/2023 05:57:15",
      "content": "<p>I guess there is nothing wrong with that. You are not expected to train the model each time you perform a submission. In other competitions, it is common practice to have train and inference notebooks. The first is for model training and the second is for the submission pretrained model. Since in most competitions, there is some time limit to submission you are likely to break this limit if you train the model right in submission.<br>\nPlease, correct me, anyone, if I am mistaking here</p>",
      "rawMarkdown": "I guess there is nothing wrong with that. You are not expected to train the model each time you perform a submission. In other competitions, it is common practice to have train and inference notebooks. The first is for model training and the second is for the submission pretrained model. Since in most competitions, there is some time limit to submission you are likely to break this limit if you train the model right in submission.\nPlease, correct me, anyone, if I am mistaking here",
      "votes": null
    },
    {
      "id": "2183373",
      "postDate": "03/15/2023 16:00:59",
      "content": "<p>That should be fine! Unless explicitly stated in the project requirements, there shouldn't be an issue, especially if your notebook doesn't require internet access. If the competition allows fine-tuning of pretrained models hosted on Kaggle I don't see why uploading your <em>own</em> model would be cause for concern…</p>",
      "rawMarkdown": "That should be fine! Unless explicitly stated in the project requirements, there shouldn't be an issue, especially if your notebook doesn't require internet access. If the competition allows fine-tuning of pretrained models hosted on Kaggle I don't see why uploading your *own* model would be cause for concern...",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2182477,
      "author_name": "meowmeowmeowmeowmeow",
      "author_url": "",
      "post_date": "03/15/2023 05:57:15",
      "content": "<p>I guess there is nothing wrong with that. You are not expected to train the model each time you perform a submission. In other competitions, it is common practice to have train and inference notebooks. The first is for model training and the second is for the submission pretrained model. Since in most competitions, there is some time limit to submission you are likely to break this limit if you train the model right in submission.<br>\nPlease, correct me, anyone, if I am mistaking here</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2183373,
      "author_name": "haydenismith",
      "author_url": "",
      "post_date": "03/15/2023 16:00:59",
      "content": "<p>That should be fine! Unless explicitly stated in the project requirements, there shouldn't be an issue, especially if your notebook doesn't require internet access. If the competition allows fine-tuning of pretrained models hosted on Kaggle I don't see why uploading your <em>own</em> model would be cause for concern…</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2182358": "Hi, I'm new to kaggle competitions so I'm not sure if this would be allowed. Can I train a model locally on my computer, and upload the trained model as a zipped file via the Kaggle dataset upload option on the notebook. I've read this as a hack to upload our models to the kaggle notebook, but I'm not sure if that is allowed in competitions. Would appreciate some help on this!",
    "2182477": "I guess there is nothing wrong with that. You are not expected to train the model each time you perform a submission. In other competitions, it is common practice to have train and inference notebooks. The first is for model training and the second is for the submission pretrained model. Since in most competitions, there is some time limit to submission you are likely to break this limit if you train the model right in submission.\nPlease, correct me, anyone, if I am mistaking here",
    "2183373": "That should be fine! Unless explicitly stated in the project requirements, there shouldn't be an issue, especially if your notebook doesn't require internet access. If the competition allows fine-tuning of pretrained models hosted on Kaggle I don't see why uploading your *own* model would be cause for concern..."
  },
  "source": "meta"
}