{
  "id": 213274,
  "title": "On hyper-parameter optimization ",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/213274",
  "author_name": "",
  "post_date": "2021-01-22T08:06:49.605061600Z",
  "votes": 5,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi guys, everyday I’m learning new stuff, and sorry for the uninformed post because I’m ignorant about the fact that there is hyperparam optimization tool. </p>\n<p>If you guys see the Jane street comp, many are using tools like Optuna etc. I am just wondering why this is not rampant in computer vision problems? I see that PyTorch Lightning has a LR finder tuner but not as exhaustive as the other packages. </p>\n<p>So is it possible to implement things like optuna, Keras tuner or Ray Tune in image classification problem? Or is it too resource intensive. </p>",
  "messages": [
    {
      "id": "1164213",
      "postDate": "01/22/2021 08:06:49",
      "content": "<p>Hi guys, everyday I’m learning new stuff, and sorry for the uninformed post because I’m ignorant about the fact that there is hyperparam optimization tool. </p>\n<p>If you guys see the Jane street comp, many are using tools like Optuna etc. I am just wondering why this is not rampant in computer vision problems? I see that PyTorch Lightning has a LR finder tuner but not as exhaustive as the other packages. </p>\n<p>So is it possible to implement things like optuna, Keras tuner or Ray Tune in image classification problem? Or is it too resource intensive. </p>",
      "rawMarkdown": "Hi guys, everyday I’m learning new stuff, and sorry for the uninformed post because I’m ignorant about the fact that there is hyperparam optimization tool. \n\nIf you guys see the Jane street comp, many are using tools like Optuna etc. I am just wondering why this is not rampant in computer vision problems? I see that PyTorch Lightning has a LR finder tuner but not as exhaustive as the other packages. \n\nSo is it possible to implement things like optuna, Keras tuner or Ray Tune in image classification problem? Or is it too resource intensive.",
      "votes": null
    },
    {
      "id": "1166434",
      "postDate": "01/23/2021 16:12:59",
      "content": "<p>The tools that you have mentioned are highly customizable. So you can use them for whichever task you want. It depends on your understanding about those tool. <a href=\"https://www.kaggle.com/reighns\" target=\"_blank\">@reighns</a> </p>",
      "rawMarkdown": "The tools that you have mentioned are highly customizable. So you can use them for whichever task you want. It depends on your understanding about those tool. @reighns",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1166434,
      "author_name": "saurabhshahane",
      "author_url": "",
      "post_date": "01/23/2021 16:12:59",
      "content": "<p>The tools that you have mentioned are highly customizable. So you can use them for whichever task you want. It depends on your understanding about those tool. <a href=\"https://www.kaggle.com/reighns\" target=\"_blank\">@reighns</a> </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1164213": "Hi guys, everyday I’m learning new stuff, and sorry for the uninformed post because I’m ignorant about the fact that there is hyperparam optimization tool. \n\nIf you guys see the Jane street comp, many are using tools like Optuna etc. I am just wondering why this is not rampant in computer vision problems? I see that PyTorch Lightning has a LR finder tuner but not as exhaustive as the other packages. \n\nSo is it possible to implement things like optuna, Keras tuner or Ray Tune in image classification problem? Or is it too resource intensive.",
    "1166434": "The tools that you have mentioned are highly customizable. So you can use them for whichever task you want. It depends on your understanding about those tool. @reighns"
  },
  "source": "meta"
}