{
  "id": 472033,
  "title": "Does efficientnet + unet + Diceloss work well? ",
  "url": "/competitions/blood-vessel-segmentation/discussion/472033",
  "author_name": "",
  "post_date": "2024-01-30T14:52:02.521645200Z",
  "votes": 3,
  "comment_count": 4,
  "views": 0,
  "content": "<p>It really slowly training or not working well… Anyone did it before?</p>",
  "messages": [
    {
      "id": "2627236",
      "postDate": "01/30/2024 14:52:02",
      "content": "<p>It really slowly training or not working well… Anyone did it before?</p>",
      "rawMarkdown": "It really slowly training or not working well... Anyone did it before?",
      "votes": null
    },
    {
      "id": "2628352",
      "postDate": "01/31/2024 08:46:57",
      "content": "<p>efficientnet + unet can yield a comparable score with popular se-resnext based method, if you ensemble this two method, you may boost you score.😀</p>",
      "rawMarkdown": "efficientnet + unet can yield a comparable score with popular se-resnext based method, if you ensemble this two method, you may boost you score.😀",
      "votes": null
    },
    {
      "id": "2628491",
      "postDate": "01/31/2024 10:32:44",
      "content": "<p>Bce loss train well then dice in my case. I don't know what loss i working with.     </p>",
      "rawMarkdown": "Bce loss train well then dice in my case. I don't know what loss i working with.",
      "votes": null
    },
    {
      "id": "2628554",
      "postDate": "01/31/2024 11:27:02",
      "content": "<p><a href=\"https://www.kaggle.com/tanxxx\" target=\"_blank\">@tanxxx</a> what image size are you using for effnet-unet?</p>",
      "rawMarkdown": "tanxxx what image size are you using for effnet-unet?",
      "votes": null
    },
    {
      "id": "2629728",
      "postDate": "02/01/2024 01:55:56",
      "content": "<p>both train and test are 1024*1024</p>",
      "rawMarkdown": "both train and test are 1024*1024",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2628352,
      "author_name": "tanxxx",
      "author_url": "",
      "post_date": "01/31/2024 08:46:57",
      "content": "<p>efficientnet + unet can yield a comparable score with popular se-resnext based method, if you ensemble this two method, you may boost you score.😀</p>",
      "votes": null,
      "replies": [
        {
          "id": 2628491,
          "author_name": "junhanzangai",
          "author_url": "",
          "post_date": "01/31/2024 10:32:44",
          "content": "<p>Bce loss train well then dice in my case. I don't know what loss i working with.     </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 2628554,
          "author_name": "projdev",
          "author_url": "",
          "post_date": "01/31/2024 11:27:02",
          "content": "<p><a href=\"https://www.kaggle.com/tanxxx\" target=\"_blank\">@tanxxx</a> what image size are you using for effnet-unet?</p>",
          "votes": null,
          "replies": [
            {
              "id": 2629728,
              "author_name": "tanxxx",
              "author_url": "",
              "post_date": "02/01/2024 01:55:56",
              "content": "<p>both train and test are 1024*1024</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2627236": "It really slowly training or not working well... Anyone did it before?",
    "2628352": "efficientnet + unet can yield a comparable score with popular se-resnext based method, if you ensemble this two method, you may boost you score.😀",
    "2628491": "Bce loss train well then dice in my case. I don't know what loss i working with.",
    "2628554": "tanxxx what image size are you using for effnet-unet?",
    "2629728": "both train and test are 1024*1024"
  },
  "source": "meta"
}