{
  "id": 428034,
  "title": "Possible shake-up?",
  "url": "/competitions/hubmap-hacking-the-human-vasculature/discussion/428034",
  "author_name": "Araik Tamazian",
  "post_date": "2023-07-30T19:47:45.221000",
  "votes": 9,
  "comment_count": 15,
  "views": 0,
  "content": "<p>Do you think will we have a shake-up here? I think there would be one, since CV-LB correlation is not very good + small size of public LB test set (just ~28% from 650 images).</p>",
  "messages": [
    {
      "id": 2366204,
      "postDate": "2023-07-30T19:53:21.293Z",
      "content": "<p>based on my experiments. I think there will be atleast +- 0.03 change of private scores if there is large domain shift in WSI 5. </p>",
      "rawMarkdown": "based on my experiments. I think there will be atleast +- 0.03 change of private scores if there is large domain shift in WSI 5. ",
      "votes": 10
    },
    {
      "id": 2366188,
      "postDate": "2023-07-30T19:47:45.223Z",
      "content": "<p>Do you think will we have a shake-up here? I think there would be one, since CV-LB correlation is not very good + small size of public LB test set (just ~28% from 650 images).</p>",
      "rawMarkdown": "Do you think will we have a shake-up here? I think there would be one, since CV-LB correlation is not very good + small size of public LB test set (just ~28% from 650 images).",
      "votes": 8
    },
    {
      "id": 2366496,
      "postDate": "2023-07-31T03:39:37.953Z",
      "content": "<p>based on my wild guess. I think there will be -0.1 change of private scores if there is large domain shift in WSI 5.</p>",
      "rawMarkdown": "based on my wild guess. I think there will be -0.1 change of private scores if there is large domain shift in WSI 5.",
      "votes": 4,
      "replies": [
        {
          "id": 2366526,
          "postDate": "2023-07-31T04:09:00.363Z",
          "content": "<p>Congrats, I noticed you went from Silver to Gold in the last week!</p>",
          "rawMarkdown": "Congrats, I noticed you went from Silver to Gold in the last week!",
          "votes": 1,
          "replies": [
            {
              "id": 2367253,
              "postDate": "2023-07-31T13:02:18.963Z",
              "content": "<p>Thanks, by the way I voted in favor for your topic \"Is someone hiding the score?\" before</p>",
              "rawMarkdown": "Thanks, by the way I voted in favor for your topic \"Is someone hiding the score?\" before"
            },
            {
              "id": 2368064,
              "postDate": "2023-08-01T02:03:27.680Z",
              "content": "<p>But I didn't expect it to be more than -0.2 😅   my final submit has lb: 0.528 and pb: 0.587 </p>",
              "rawMarkdown": "But I didn't expect it to be more than -0.2 😅   my final submit has lb: 0.528 and pb: 0.587 "
            },
            {
              "id": 2368070,
              "postDate": "2023-08-01T02:06:52.170Z",
              "content": "<p>你是提交了带有dilate的版本？</p>",
              "rawMarkdown": "你是提交了带有dilate的版本？"
            },
            {
              "id": 2368136,
              "postDate": "2023-08-01T02:59:05.997Z",
              "content": "<p>这比赛真的是坐牢hhh</p>",
              "rawMarkdown": "这比赛真的是坐牢hhh",
              "votes": 1
            },
            {
              "id": 2368490,
              "postDate": "2023-08-01T07:45:12.270Z",
              "content": "<p>I have a silver-medal worth submission with private LB of 0.452 (no dilation), but I didn't chose it since its public LB was just 0.343 😔</p>",
              "rawMarkdown": "I have a silver-medal worth submission with private LB of 0.452 (no dilation), but I didn't chose it since its public LB was just 0.343 😔"
            },
            {
              "id": 2368533,
              "postDate": "2023-08-01T08:17:44.017Z",
              "content": "<p>是的，我又做了一些实验，dilate是罪魁祸首</p>",
              "rawMarkdown": "是的，我又做了一些实验，dilate是罪魁祸首"
            }
          ]
        },
        {
          "id": 2370258,
          "postDate": "2023-08-02T09:19:56.873Z",
          "rawMarkdown": "",
          "isDeleted": true
        }
      ]
    },
    {
      "id": 2367492,
      "postDate": "2023-07-31T15:23:18.020Z",
      "content": "<p>Lucky is all I need, dino self-supervised with external data isn't work with me -_-.</p>",
      "rawMarkdown": "Lucky is all I need, dino self-supervised with external data isn't work with me -_-."
    },
    {
      "id": 2367463,
      "postDate": "2023-07-31T15:05:34.660Z",
      "content": "<p>Also, that issue with dilation… I wonder how it will turn out. I guess nearly everyone in LB top is using it.</p>",
      "rawMarkdown": "Also, that issue with dilation... I wonder how it will turn out. I guess nearly everyone in LB top is using it.",
      "replies": [
        {
          "id": 2367498,
          "postDate": "2023-07-31T15:30:22.470Z",
          "rawMarkdown": "",
          "isDeleted": true
        },
        {
          "id": 2368042,
          "postDate": "2023-08-01T01:46:28.740Z",
          "content": "<p>You are a seer!</p>",
          "rawMarkdown": "You are a seer!",
          "votes": 1
        },
        {
          "id": 2368045,
          "postDate": "2023-08-01T01:48:25.107Z",
          "content": "<p>I was struggling for a long time yesterday because I read this reply of yours, but in the end I didn't choose non-dilate</p>",
          "rawMarkdown": "I was struggling for a long time yesterday because I read this reply of yours, but in the end I didn't choose non-dilate"
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 2366204,
      "author_name": "Nischay Dhankhar",
      "author_url": "",
      "post_date": "2023-07-30T19:53:21.293000",
      "content": "<p>based on my experiments. I think there will be atleast +- 0.03 change of private scores if there is large domain shift in WSI 5. </p>",
      "votes": 10,
      "replies": []
    },
    {
      "id": 2366496,
      "author_name": "clwclw",
      "author_url": "",
      "post_date": "2023-07-31T03:39:37.953000",
      "content": "<p>based on my wild guess. I think there will be -0.1 change of private scores if there is large domain shift in WSI 5.</p>",
      "votes": 4,
      "replies": [
        {
          "id": 2366526,
          "author_name": "bent1e",
          "author_url": "",
          "post_date": "2023-07-31T04:09:00.363000",
          "content": "<p>Congrats, I noticed you went from Silver to Gold in the last week!</p>",
          "votes": 1,
          "replies": [
            {
              "id": 2367253,
              "author_name": "clwclw",
              "author_url": "",
              "post_date": "2023-07-31T13:02:18.963000",
              "content": "<p>Thanks, by the way I voted in favor for your topic \"Is someone hiding the score?\" before</p>",
              "votes": 0,
              "replies": []
            },
            {
              "id": 2368064,
              "author_name": "clwclw",
              "author_url": "",
              "post_date": "2023-08-01T02:03:27.680000",
              "content": "<p>But I didn't expect it to be more than -0.2 😅   my final submit has lb: 0.528 and pb: 0.587 </p>",
              "votes": 0,
              "replies": []
            },
            {
              "id": 2368070,
              "author_name": "bent1e",
              "author_url": "",
              "post_date": "2023-08-01T02:06:52.170000",
              "content": "<p>你是提交了带有dilate的版本？</p>",
              "votes": 0,
              "replies": []
            },
            {
              "id": 2368136,
              "author_name": "RickyLu",
              "author_url": "",
              "post_date": "2023-08-01T02:59:05.997000",
              "content": "<p>这比赛真的是坐牢hhh</p>",
              "votes": 1,
              "replies": []
            },
            {
              "id": 2368490,
              "author_name": "Araik Tamazian",
              "author_url": "",
              "post_date": "2023-08-01T07:45:12.270000",
              "content": "<p>I have a silver-medal worth submission with private LB of 0.452 (no dilation), but I didn't chose it since its public LB was just 0.343 😔</p>",
              "votes": 0,
              "replies": []
            },
            {
              "id": 2368533,
              "author_name": "clwclw",
              "author_url": "",
              "post_date": "2023-08-01T08:17:44.017000",
              "content": "<p>是的，我又做了一些实验，dilate是罪魁祸首</p>",
              "votes": 0,
              "replies": []
            }
          ]
        },
        {
          "id": 2370258,
          "author_name": "",
          "author_url": "",
          "post_date": "2023-08-02T09:19:56.873000",
          "content": "",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 2367492,
      "author_name": "Quan Vu",
      "author_url": "",
      "post_date": "2023-07-31T15:23:18.020000",
      "content": "<p>Lucky is all I need, dino self-supervised with external data isn't work with me -_-.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 2367463,
      "author_name": "Araik Tamazian",
      "author_url": "",
      "post_date": "2023-07-31T15:05:34.660000",
      "content": "<p>Also, that issue with dilation… I wonder how it will turn out. I guess nearly everyone in LB top is using it.</p>",
      "votes": 0,
      "replies": [
        {
          "id": 2367498,
          "author_name": "",
          "author_url": "",
          "post_date": "2023-07-31T15:30:22.470000",
          "content": "",
          "votes": 0,
          "replies": []
        },
        {
          "id": 2368042,
          "author_name": "RickyLu",
          "author_url": "",
          "post_date": "2023-08-01T01:46:28.740000",
          "content": "<p>You are a seer!</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 2368045,
          "author_name": "bent1e",
          "author_url": "",
          "post_date": "2023-08-01T01:48:25.107000",
          "content": "<p>I was struggling for a long time yesterday because I read this reply of yours, but in the end I didn't choose non-dilate</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2366204": "based on my experiments. I think there will be atleast +- 0.03 change of private scores if there is large domain shift in WSI 5. ",
    "2366188": "Do you think will we have a shake-up here? I think there would be one, since CV-LB correlation is not very good + small size of public LB test set (just ~28% from 650 images).",
    "2366496": "based on my wild guess. I think there will be -0.1 change of private scores if there is large domain shift in WSI 5.",
    "2367492": "Lucky is all I need, dino self-supervised with external data isn't work with me -_-.",
    "2367463": "Also, that issue with dilation... I wonder how it will turn out. I guess nearly everyone in LB top is using it."
  }
}