{
  "id": 428960,
  "title": "There are something we need to learn from this competition",
  "url": "/competitions/hubmap-hacking-the-human-vasculature/discussion/428960",
  "author_name": "Keli",
  "post_date": "2023-08-03T14:10:06.403000",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<ul>\n<li>All of the test set tiles are from Dataset 1.</li>\n<li>Two of the WSIs make up the training set, two WSIs make up the public test set, and one WSI makes up the private test set.</li>\n<li>The training data includes Dataset 2 tiles from the <em>public</em> test WSI, but <em>not</em> from the <em>private</em> test WSI.</li>\n</ul>\n<p>At first I was very angry when I saw the huge shake in the LB, but after thinking about it, I got some sense.<br>\nWe need to know why we're playing the game, and what we need to build is a general model that works well across different wsi's. We should not be concerned about how to make our private LB higher, we should be concerned about how to make both public LB and private LB higher (because they are from different WSI), in this regard I believe someone has done this.</p>\n<p>I hope everyone  can share and discuss the  solutions which can get high score in public LB, private LB and CV</p>",
  "messages": [
    {
      "id": 2372145,
      "postDate": "2023-08-03T14:10:06.403Z",
      "content": "<ul>\n<li>All of the test set tiles are from Dataset 1.</li>\n<li>Two of the WSIs make up the training set, two WSIs make up the public test set, and one WSI makes up the private test set.</li>\n<li>The training data includes Dataset 2 tiles from the <em>public</em> test WSI, but <em>not</em> from the <em>private</em> test WSI.</li>\n</ul>\n<p>At first I was very angry when I saw the huge shake in the LB, but after thinking about it, I got some sense.<br>\nWe need to know why we're playing the game, and what we need to build is a general model that works well across different wsi's. We should not be concerned about how to make our private LB higher, we should be concerned about how to make both public LB and private LB higher (because they are from different WSI), in this regard I believe someone has done this.</p>\n<p>I hope everyone  can share and discuss the  solutions which can get high score in public LB, private LB and CV</p>",
      "rawMarkdown": "- All of the test set tiles are from Dataset 1.\n- Two of the WSIs make up the training set, two WSIs make up the public test set, and one WSI makes up the private test set.\n- The training data includes Dataset 2 tiles from the *public* test WSI, but *not* from the *private* test WSI.\n\nAt first I was very angry when I saw the huge shake in the LB, but after thinking about it, I got some sense.\nWe need to know why we're playing the game, and what we need to build is a general model that works well across different wsi's. We should not be concerned about how to make our private LB higher, we should be concerned about how to make both public LB and private LB higher (because they are from different WSI), in this regard I believe someone has done this.\n\nI hope everyone  can share and discuss the  solutions which can get high score in public LB, private LB and CV",
      "votes": 2
    },
    {
      "id": 2372843,
      "postDate": "2023-08-04T02:13:46.917Z",
      "content": "<p>For newbies like me, it's more practical to build good local cross-validation, so I can at least keep a bronze meda.</p>",
      "rawMarkdown": "For newbies like me, it's more practical to build good local cross-validation, so I can at least keep a bronze meda.",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 2372843,
      "author_name": "",
      "author_url": "",
      "post_date": "2023-08-04T02:13:46.917000",
      "content": "<p>For newbies like me, it's more practical to build good local cross-validation, so I can at least keep a bronze meda.</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2372145": "- All of the test set tiles are from Dataset 1.\n- Two of the WSIs make up the training set, two WSIs make up the public test set, and one WSI makes up the private test set.\n- The training data includes Dataset 2 tiles from the *public* test WSI, but *not* from the *private* test WSI.\n\nAt first I was very angry when I saw the huge shake in the LB, but after thinking about it, I got some sense.\nWe need to know why we're playing the game, and what we need to build is a general model that works well across different wsi's. We should not be concerned about how to make our private LB higher, we should be concerned about how to make both public LB and private LB higher (because they are from different WSI), in this regard I believe someone has done this.\n\nI hope everyone  can share and discuss the  solutions which can get high score in public LB, private LB and CV",
    "2372843": "For newbies like me, it's more practical to build good local cross-validation, so I can at least keep a bronze meda."
  }
}