{
  "id": 101768,
  "title": "Training from Scratch",
  "url": "/competitions/aptos2019-blindness-detection/discussion/101768",
  "author_name": "",
  "post_date": "2019-07-28T13:53:22.071069400Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello, Kagglers. \nIt seems like almost all use pre-trained models for this competition. I'm curious if someone trains any model from scratch and does this model have comparable results?</p>",
  "messages": [
    {
      "id": "586049",
      "postDate": "07/28/2019 13:53:22",
      "content": "<p>Hello, Kagglers. \nIt seems like almost all use pre-trained models for this competition. I'm curious if someone trains any model from scratch and does this model have comparable results?</p>",
      "rawMarkdown": "Hello, Kagglers. \nIt seems like almost all use pre-trained models for this competition. I'm curious if someone trains any model from scratch and does this model have comparable results?",
      "votes": null
    },
    {
      "id": "590736",
      "postDate": "08/02/2019 14:25:35",
      "content": "<p>Training from scratch did not go well for me.  I think there's too less data to train from scratch.</p>",
      "rawMarkdown": "Training from scratch did not go well for me.  I think there's too less data to train from scratch.",
      "votes": null
    },
    {
      "id": "591342",
      "postDate": "08/03/2019 14:24:14",
      "content": "<p>Thanks</p>",
      "rawMarkdown": "Thanks",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 590736,
      "author_name": "harshthaker",
      "author_url": "",
      "post_date": "08/02/2019 14:25:35",
      "content": "<p>Training from scratch did not go well for me.  I think there's too less data to train from scratch.</p>",
      "votes": null,
      "replies": [
        {
          "id": 591342,
          "author_name": "ademyanchuk",
          "author_url": "",
          "post_date": "08/03/2019 14:24:14",
          "content": "<p>Thanks</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "586049": "Hello, Kagglers. \nIt seems like almost all use pre-trained models for this competition. I'm curious if someone trains any model from scratch and does this model have comparable results?",
    "590736": "Training from scratch did not go well for me.  I think there's too less data to train from scratch.",
    "591342": "Thanks"
  },
  "source": "meta"
}