{
  "id": 469451,
  "title": "Hyperparameter tuning allowed?",
  "url": "/competitions/bench-tab-v1/discussion/469451",
  "author_name": "",
  "post_date": "2024-01-20T16:12:29.545562Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>The question might sound silly, but I wonder if hyperparameter tuning is allowed in this competition. </p>\n<p>Are we trying to find the best architecture or the best model on default params?</p>",
  "messages": [
    {
      "id": "2611221",
      "postDate": "01/20/2024 16:12:29",
      "content": "<p>The question might sound silly, but I wonder if hyperparameter tuning is allowed in this competition. </p>\n<p>Are we trying to find the best architecture or the best model on default params?</p>",
      "rawMarkdown": "The question might sound silly, but I wonder if hyperparameter tuning is allowed in this competition. \n\nAre we trying to find the best architecture or the best model on default params?",
      "votes": null
    },
    {
      "id": "2611314",
      "postDate": "01/20/2024 17:08:01",
      "content": "<p>Yes it is allowed, I don't see why it wouldn't be. Plus, in a lot of NN cases there isn't such a thing as 'default parameters'.</p>",
      "rawMarkdown": "Yes it is allowed, I don't see why it wouldn't be. Plus, in a lot of NN cases there isn't such a thing as 'default parameters'.",
      "votes": null
    },
    {
      "id": "2611358",
      "postDate": "01/20/2024 17:26:39",
      "content": "<p>It will be allowed for sure <a href=\"https://www.kaggle.com/thomasmeiner\" target=\"_blank\">@thomasmeiner</a> <br>\nDefault parameters may sometimes not work, here we have 45 datasets, so tuning could be a task in itself!!!</p>",
      "rawMarkdown": "It will be allowed for sure @thomasmeiner \nDefault parameters may sometimes not work, here we have 45 datasets, so tuning could be a task in itself!!!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2611314,
      "author_name": "lucasmorin",
      "author_url": "",
      "post_date": "01/20/2024 17:08:01",
      "content": "<p>Yes it is allowed, I don't see why it wouldn't be. Plus, in a lot of NN cases there isn't such a thing as 'default parameters'.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2611358,
      "author_name": "ravi20076",
      "author_url": "",
      "post_date": "01/20/2024 17:26:39",
      "content": "<p>It will be allowed for sure <a href=\"https://www.kaggle.com/thomasmeiner\" target=\"_blank\">@thomasmeiner</a> <br>\nDefault parameters may sometimes not work, here we have 45 datasets, so tuning could be a task in itself!!!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2611221": "The question might sound silly, but I wonder if hyperparameter tuning is allowed in this competition. \n\nAre we trying to find the best architecture or the best model on default params?",
    "2611314": "Yes it is allowed, I don't see why it wouldn't be. Plus, in a lot of NN cases there isn't such a thing as 'default parameters'.",
    "2611358": "It will be allowed for sure @thomasmeiner \nDefault parameters may sometimes not work, here we have 45 datasets, so tuning could be a task in itself!!!"
  },
  "source": "meta"
}