{
  "id": 260948,
  "title": "2d efficientnet or 3d efficientnet?",
  "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/260948",
  "author_name": "",
  "post_date": "2021-08-03T23:24:01.464645600Z",
  "votes": 8,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Completely new to this competition.<br>\nI realized that there are two popular approaches in this competition: 2d efficientnet and 3d efficientnet. <br>\nAre there any benefits to choosing one over another?</p>\n<p>Thank you.</p>",
  "messages": [
    {
      "id": "1442569",
      "postDate": "08/03/2021 23:24:01",
      "content": "<p>Completely new to this competition.<br>\nI realized that there are two popular approaches in this competition: 2d efficientnet and 3d efficientnet. <br>\nAre there any benefits to choosing one over another?</p>\n<p>Thank you.</p>",
      "rawMarkdown": "Completely new to this competition.\nI realized that there are two popular approaches in this competition: 2d efficientnet and 3d efficientnet. \nAre there any benefits to choosing one over another?\n\nThank you.",
      "votes": null
    },
    {
      "id": "1444934",
      "postDate": "08/04/2021 07:22:56",
      "content": "<p>3d models need more time to train, it's computationally expensive.<br>\nNot a problem if you have a  good GPU though.</p>",
      "rawMarkdown": "3d models need more time to train, it's computationally expensive.\nNot a problem if you have a  good GPU though.",
      "votes": null
    },
    {
      "id": "1524247",
      "postDate": "09/26/2021 09:43:03",
      "content": "<p>I think 3D is better, 2D may get overfitting, but 3D may help you, this is my experience on this data </p>",
      "rawMarkdown": "I think 3D is better, 2D may get overfitting, but 3D may help you, this is my experience on this data",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1444934,
      "author_name": "mrinath",
      "author_url": "",
      "post_date": "08/04/2021 07:22:56",
      "content": "<p>3d models need more time to train, it's computationally expensive.<br>\nNot a problem if you have a  good GPU though.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1524247,
      "author_name": "huyquoctrinh",
      "author_url": "",
      "post_date": "09/26/2021 09:43:03",
      "content": "<p>I think 3D is better, 2D may get overfitting, but 3D may help you, this is my experience on this data </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1442569": "Completely new to this competition.\nI realized that there are two popular approaches in this competition: 2d efficientnet and 3d efficientnet. \nAre there any benefits to choosing one over another?\n\nThank you.",
    "1444934": "3d models need more time to train, it's computationally expensive.\nNot a problem if you have a  good GPU though.",
    "1524247": "I think 3D is better, 2D may get overfitting, but 3D may help you, this is my experience on this data"
  },
  "source": "meta"
}