{
  "id": 238034,
  "title": "single model gold medal(private 948) solution",
  "url": "/competitions/hubmap-kidney-segmentation/writeups/only-trust-cv-single-model-gold-medal-private-948-",
  "author_name": "",
  "post_date": "2021-05-11T02:40:42.980851600Z",
  "votes": 5,
  "comment_count": 2,
  "views": 0,
  "content": "<p>It's hard to say it's a happy journey. <br>\nWe missed the private 948 solution submitted last night.😄<br>\n<img src=\"https://i.loli.net/2021/05/11/Zfyk6rX7MHCRdv4.png\" alt=\"微信截图_20210511090244.png\"><br>\nAnd I found that an experiment with lb912 got private 948, which I trained a month ago.<br>\n<img src=\"https://i.loli.net/2021/05/11/Em2K3QYlXrOTwjs.png\" alt=\"微信截图_20210511090219.png\"><br>\nI share the train, inference and tiles-making code of exp007 here.<br>\ntrain: <a href=\"https://www.kaggle.com/tiandaye/hubmap-exp007-train?scriptVersionId=62523292\" target=\"_blank\">https://www.kaggle.com/tiandaye/hubmap-exp007-train?scriptVersionId=62523292</a><br>\ninference: <a href=\"https://www.kaggle.com/tiandaye/hubmap-exp007-5folds?scriptVersionId=58430396\" target=\"_blank\">https://www.kaggle.com/tiandaye/hubmap-exp007-5folds?scriptVersionId=58430396</a><br>\ntiles making: <a href=\"https://www.kaggle.com/tiandaye/make-tiles\" target=\"_blank\">https://www.kaggle.com/tiandaye/make-tiles</a></p>\n<p>Unet with encoder='timm-efficientnet-b4' , weights='noisy-student', trained with label smooth and no PL.</p>",
  "messages": [
    {
      "id": "1301303",
      "postDate": "05/11/2021 02:40:42",
      "content": "<p>It's hard to say it's a happy journey. <br>\nWe missed the private 948 solution submitted last night.😄<br>\n<img src=\"https://i.loli.net/2021/05/11/Zfyk6rX7MHCRdv4.png\" alt=\"微信截图_20210511090244.png\"><br>\nAnd I found that an experiment with lb912 got private 948, which I trained a month ago.<br>\n<img src=\"https://i.loli.net/2021/05/11/Em2K3QYlXrOTwjs.png\" alt=\"微信截图_20210511090219.png\"><br>\nI share the train, inference and tiles-making code of exp007 here.<br>\ntrain: <a href=\"https://www.kaggle.com/tiandaye/hubmap-exp007-train?scriptVersionId=62523292\" target=\"_blank\">https://www.kaggle.com/tiandaye/hubmap-exp007-train?scriptVersionId=62523292</a><br>\ninference: <a href=\"https://www.kaggle.com/tiandaye/hubmap-exp007-5folds?scriptVersionId=58430396\" target=\"_blank\">https://www.kaggle.com/tiandaye/hubmap-exp007-5folds?scriptVersionId=58430396</a><br>\ntiles making: <a href=\"https://www.kaggle.com/tiandaye/make-tiles\" target=\"_blank\">https://www.kaggle.com/tiandaye/make-tiles</a></p>\n<p>Unet with encoder='timm-efficientnet-b4' , weights='noisy-student', trained with label smooth and no PL.</p>",
      "rawMarkdown": "It's hard to say it's a happy journey. \nWe missed the private 948 solution submitted last night.😄\n![微信截图_20210511090244.png](https://i.loli.net/2021/05/11/Zfyk6rX7MHCRdv4.png)\nAnd I found that an experiment with lb912 got private 948, which I trained a month ago.\n![微信截图_20210511090219.png](https://i.loli.net/2021/05/11/Em2K3QYlXrOTwjs.png)\nI share the train, inference and tiles-making code of exp007 here.\ntrain: [https://www.kaggle.com/tiandaye/hubmap-exp007-train?scriptVersionId=62523292](https://www.kaggle.com/tiandaye/hubmap-exp007-train?scriptVersionId=62523292)\ninference: [https://www.kaggle.com/tiandaye/hubmap-exp007-5folds?scriptVersionId=58430396](https://www.kaggle.com/tiandaye/hubmap-exp007-5folds?scriptVersionId=58430396)\ntiles making: [https://www.kaggle.com/tiandaye/make-tiles](https://www.kaggle.com/tiandaye/make-tiles)\n\nUnet with encoder='timm-efficientnet-b4' , weights='noisy-student', trained with label smooth and no PL.",
      "votes": null
    },
    {
      "id": "1301397",
      "postDate": "05/11/2021 04:08:28",
      "content": "<p>My Unetpp-effb6ns model also got 0.948 haha</p>",
      "rawMarkdown": "My Unetpp-effb6ns model also got 0.948 haha",
      "votes": null
    },
    {
      "id": "1303052",
      "postDate": "05/11/2021 21:50:20",
      "content": "<p>Pity party 🤜</p>",
      "rawMarkdown": "Pity party 🤜",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1301397,
      "author_name": "yienngxiong",
      "author_url": "",
      "post_date": "05/11/2021 04:08:28",
      "content": "<p>My Unetpp-effb6ns model also got 0.948 haha</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1303052,
      "author_name": "underwearfitting",
      "author_url": "",
      "post_date": "05/11/2021 21:50:20",
      "content": "<p>Pity party 🤜</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1301303": "It's hard to say it's a happy journey. \nWe missed the private 948 solution submitted last night.😄\n![微信截图_20210511090244.png](https://i.loli.net/2021/05/11/Zfyk6rX7MHCRdv4.png)\nAnd I found that an experiment with lb912 got private 948, which I trained a month ago.\n![微信截图_20210511090219.png](https://i.loli.net/2021/05/11/Em2K3QYlXrOTwjs.png)\nI share the train, inference and tiles-making code of exp007 here.\ntrain: [https://www.kaggle.com/tiandaye/hubmap-exp007-train?scriptVersionId=62523292](https://www.kaggle.com/tiandaye/hubmap-exp007-train?scriptVersionId=62523292)\ninference: [https://www.kaggle.com/tiandaye/hubmap-exp007-5folds?scriptVersionId=58430396](https://www.kaggle.com/tiandaye/hubmap-exp007-5folds?scriptVersionId=58430396)\ntiles making: [https://www.kaggle.com/tiandaye/make-tiles](https://www.kaggle.com/tiandaye/make-tiles)\n\nUnet with encoder='timm-efficientnet-b4' , weights='noisy-student', trained with label smooth and no PL.",
    "1301397": "My Unetpp-effb6ns model also got 0.948 haha",
    "1303052": "Pity party 🤜"
  },
  "source": "meta"
}