{
  "id": 353428,
  "title": "How to better integrate our model?",
  "url": "/competitions/hubmap-organ-segmentation/discussion/353428",
  "author_name": "",
  "post_date": "2022-09-18T11:29:16.320680100Z",
  "votes": 6,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hello everyone, when we fuse the models, we often use the cumulative average to obtain the final Rle, but if the models are fused too much, there will be great drawbacks.<br>\n For example, we may not know which models conflict with each other and which models are redundant in the entire fusion set, which may result in lower scores or longer submission times.<br>\n Are there better suggestions to make the models blend more smoothly?🙏🙏🙏</p>",
  "messages": [
    {
      "id": "1944553",
      "postDate": "09/18/2022 11:29:16",
      "content": "<p>Hello everyone, when we fuse the models, we often use the cumulative average to obtain the final Rle, but if the models are fused too much, there will be great drawbacks.<br>\n For example, we may not know which models conflict with each other and which models are redundant in the entire fusion set, which may result in lower scores or longer submission times.<br>\n Are there better suggestions to make the models blend more smoothly?🙏🙏🙏</p>",
      "rawMarkdown": "Hello everyone, when we fuse the models, we often use the cumulative average to obtain the final Rle, but if the models are fused too much, there will be great drawbacks.\n For example, we may not know which models conflict with each other and which models are redundant in the entire fusion set, which may result in lower scores or longer submission times.\n Are there better suggestions to make the models blend more smoothly?🙏🙏🙏",
      "votes": null
    },
    {
      "id": "1977347",
      "postDate": "10/08/2022 01:03:33",
      "content": "<p>Marvelous！！！🙀</p>",
      "rawMarkdown": "Marvelous！！！🙀",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1977347,
      "author_name": "srilankalt",
      "author_url": "",
      "post_date": "10/08/2022 01:03:33",
      "content": "<p>Marvelous！！！🙀</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1944553": "Hello everyone, when we fuse the models, we often use the cumulative average to obtain the final Rle, but if the models are fused too much, there will be great drawbacks.\n For example, we may not know which models conflict with each other and which models are redundant in the entire fusion set, which may result in lower scores or longer submission times.\n Are there better suggestions to make the models blend more smoothly?🙏🙏🙏",
    "1977347": "Marvelous！！！🙀"
  },
  "source": "meta"
}