{
  "id": 238012,
  "title": "Vision Transformer inference results ",
  "url": "/competitions/hubmap-kidney-segmentation/discussion/238012",
  "author_name": "Kassem",
  "post_date": "2021-05-11T01:12:33.198000",
  "votes": 4,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hello,<br>\nCongrats to all the winners and best of luck to everyone. </p>\n<p>At the beginning of the competition I have shared <strong><a href=\"https://www.kaggle.com/elcaiseri/hubmap-pytorch-vit-segmentation-starter-train\">HuBMAP| Pytorch | ViT Segmentation Starter [Train]</a></strong> notebook and a few people asked me the inference notebook. </p>\n<p>As you can see the model was too hard to handle so I trained only one fold and ensamble it with one of the public weights by <a href=\"https://www.kaggle.com/iafoss\" target=\"_blank\">@iafoss</a> in the inference mode. </p>\n<p>The results was quite impressive :</p>\n<ul>\n<li>V3</li>\n<li>Public score : 0.905</li>\n<li><strong>Private score : 0.935</strong></li>\n</ul>\n<p>You can find the inference notebook here:</p>\n<p><strong><a href=\"https://www.kaggle.com/elcaiseri/hubmap-pytorch-vit-segmentation-sub1\">HuBMAP| Pytorch | ViT Segmentation [Sub1]</a></strong></p>",
  "messages": [
    {
      "id": 1301173,
      "postDate": "2021-05-11T01:12:33.200Z",
      "content": "<p>Hello,<br>\nCongrats to all the winners and best of luck to everyone. </p>\n<p>At the beginning of the competition I have shared <strong><a href=\"https://www.kaggle.com/elcaiseri/hubmap-pytorch-vit-segmentation-starter-train\">HuBMAP| Pytorch | ViT Segmentation Starter [Train]</a></strong> notebook and a few people asked me the inference notebook. </p>\n<p>As you can see the model was too hard to handle so I trained only one fold and ensamble it with one of the public weights by <a href=\"https://www.kaggle.com/iafoss\" target=\"_blank\">@iafoss</a> in the inference mode. </p>\n<p>The results was quite impressive :</p>\n<ul>\n<li>V3</li>\n<li>Public score : 0.905</li>\n<li><strong>Private score : 0.935</strong></li>\n</ul>\n<p>You can find the inference notebook here:</p>\n<p><strong><a href=\"https://www.kaggle.com/elcaiseri/hubmap-pytorch-vit-segmentation-sub1\">HuBMAP| Pytorch | ViT Segmentation [Sub1]</a></strong></p>",
      "rawMarkdown": "Hello,\nCongrats to all the winners and best of luck to everyone. \n\nAt the beginning of the competition I have shared **<a href='https://www.kaggle.com/elcaiseri/hubmap-pytorch-vit-segmentation-starter-train' >HuBMAP| Pytorch | ViT Segmentation Starter [Train]<\\a>** notebook and a few people asked me the inference notebook. \n\nAs you can see the model was too hard to handle so I trained only one fold and ensamble it with one of the public weights by @iafoss in the inference mode. \n\nThe results was quite impressive :\n- V3\n- Public score : 0.905\n- **Private score : 0.935**\n\nYou can find the inference notebook here:\n\n**<a href='https://www.kaggle.com/elcaiseri/hubmap-pytorch-vit-segmentation-sub1' >HuBMAP| Pytorch | ViT Segmentation [Sub1]<\\a>**\n",
      "votes": 4
    },
    {
      "id": 1308672,
      "postDate": "2021-05-15T11:22:12.207Z",
      "content": "<p>hi <a href=\"https://www.kaggle.com/elcaiseri\" target=\"_blank\">@elcaiseri</a>,</p>\n<p>Thank you for sharing. This is very useful.</p>",
      "rawMarkdown": "hi @elcaiseri,\n\nThank you for sharing. This is very useful.",
      "votes": 1,
      "replies": [
        {
          "id": 1309363,
          "postDate": "2021-05-15T21:31:01.140Z",
          "content": "<p>Y are welcome! </p>",
          "rawMarkdown": "Y are welcome! "
        }
      ]
    },
    {
      "id": 1318662,
      "postDate": "2021-05-22T13:25:25.863Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 1308672,
      "author_name": "supplejade",
      "author_url": "",
      "post_date": "2021-05-15T11:22:12.207000",
      "content": "<p>hi <a href=\"https://www.kaggle.com/elcaiseri\" target=\"_blank\">@elcaiseri</a>,</p>\n<p>Thank you for sharing. This is very useful.</p>",
      "votes": 1,
      "replies": [
        {
          "id": 1309363,
          "author_name": "Kassem",
          "author_url": "",
          "post_date": "2021-05-15T21:31:01.140000",
          "content": "<p>Y are welcome! </p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 1318662,
      "author_name": "",
      "author_url": "",
      "post_date": "2021-05-22T13:25:25.863000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1301173": "Hello,\nCongrats to all the winners and best of luck to everyone. \n\nAt the beginning of the competition I have shared **<a href='https://www.kaggle.com/elcaiseri/hubmap-pytorch-vit-segmentation-starter-train' >HuBMAP| Pytorch | ViT Segmentation Starter [Train]<\\a>** notebook and a few people asked me the inference notebook. \n\nAs you can see the model was too hard to handle so I trained only one fold and ensamble it with one of the public weights by @iafoss in the inference mode. \n\nThe results was quite impressive :\n- V3\n- Public score : 0.905\n- **Private score : 0.935**\n\nYou can find the inference notebook here:\n\n**<a href='https://www.kaggle.com/elcaiseri/hubmap-pytorch-vit-segmentation-sub1' >HuBMAP| Pytorch | ViT Segmentation [Sub1]<\\a>**\n",
    "1308672": "hi @elcaiseri,\n\nThank you for sharing. This is very useful.",
    "1318662": ""
  }
}