{
  "id": 228626,
  "title": "How to implement SnapMix DataAugumentation in Fastai?",
  "url": "/competitions/plant-pathology-2021-fgvc8/discussion/228626",
  "author_name": "",
  "post_date": "2021-03-25T14:41:29.485120900Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I am finding difficulty in integrating Snapmix in Fastai, Any good source How to do it!! <br>\nThanks :)</p>",
  "messages": [
    {
      "id": "1252261",
      "postDate": "03/25/2021 14:41:29",
      "content": "<p>I am finding difficulty in integrating Snapmix in Fastai, Any good source How to do it!! <br>\nThanks :)</p>",
      "rawMarkdown": "I am finding difficulty in integrating Snapmix in Fastai, Any good source How to do it!! \nThanks :)",
      "votes": null
    },
    {
      "id": "1300268",
      "postDate": "05/10/2021 11:09:29",
      "content": "<p>To be honest, I don't think using different types of mix strategies would give you a performance boost. Typically, mixup or cutmix should be efficient enough to improve the robustness of your model.</p>",
      "rawMarkdown": "To be honest, I don't think using different types of mix strategies would give you a performance boost. Typically, mixup or cutmix should be efficient enough to improve the robustness of your model.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1300268,
      "author_name": "ytepzhi",
      "author_url": "",
      "post_date": "05/10/2021 11:09:29",
      "content": "<p>To be honest, I don't think using different types of mix strategies would give you a performance boost. Typically, mixup or cutmix should be efficient enough to improve the robustness of your model.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1252261": "I am finding difficulty in integrating Snapmix in Fastai, Any good source How to do it!! \nThanks :)",
    "1300268": "To be honest, I don't think using different types of mix strategies would give you a performance boost. Typically, mixup or cutmix should be efficient enough to improve the robustness of your model."
  },
  "source": "meta"
}