{
  "id": 446518,
  "title": " train locally without downloading data",
  "url": "/competitions/UBC-OCEAN/discussion/446518",
  "author_name": "Huijie Liu",
  "post_date": "2023-10-12T04:10:02.975000",
  "votes": 3,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Hello, I would like to ask if there is a way to train locally without downloading data for such big data topics. 🥰</p>",
  "messages": [
    {
      "id": 2481172,
      "postDate": "2023-10-13T21:44:34.507Z",
      "content": "<p>I did some preprocessing and cropping in a Kaggle notebook and downloaded the resulting patches to do all the training locally. There are already some examples in the code section, e.g. <a href=\"https://www.kaggle.com/code/jirkaborovec/cancer-subtype-decompose-large-image-tiles\" target=\"_blank\">https://www.kaggle.com/code/jirkaborovec/cancer-subtype-decompose-large-image-tiles</a><br>\nHope this helps.</p>",
      "rawMarkdown": "I did some preprocessing and cropping in a Kaggle notebook and downloaded the resulting patches to do all the training locally. There are already some examples in the code section, e.g. https://www.kaggle.com/code/jirkaborovec/cancer-subtype-decompose-large-image-tiles\nHope this helps.",
      "votes": 3,
      "replies": [
        {
          "id": 2481292,
          "postDate": "2023-10-14T02:39:18.397Z",
          "content": "<p>Thank you. Very good job.</p>",
          "rawMarkdown": "Thank you. Very good job."
        }
      ]
    },
    {
      "id": 2478512,
      "postDate": "2023-10-12T04:10:02.977Z",
      "content": "<p>Hello, I would like to ask if there is a way to train locally without downloading data for such big data topics. 🥰</p>",
      "rawMarkdown": "Hello, I would like to ask if there is a way to train locally without downloading data for such big data topics. 🥰",
      "votes": 3
    },
    {
      "id": 2478797,
      "postDate": "2023-10-12T08:04:27.743Z",
      "content": "<p>Since given GPU in Kaggle is P100 16g which it might not be able to train WSI（Whole Slide Imaging）and even you can it pretty slow I think😬，and you can download data with Kaggle API it‘s may take a night ~</p>",
      "rawMarkdown": "Since given GPU in Kaggle is P100 16g which it might not be able to train WSI（Whole Slide Imaging）and even you can it pretty slow I think😬，and you can download data with Kaggle API it‘s may take a night ~",
      "votes": 2,
      "replies": [
        {
          "id": 2479011,
          "postDate": "2023-10-12T10:41:33.563Z",
          "content": "<p>Yeah, so I think local training might be more appropriate.</p>",
          "rawMarkdown": "Yeah, so I think local training might be more appropriate.",
          "votes": 1
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 2481172,
      "author_name": "Manuel K",
      "author_url": "",
      "post_date": "2023-10-13T21:44:34.507000",
      "content": "<p>I did some preprocessing and cropping in a Kaggle notebook and downloaded the resulting patches to do all the training locally. There are already some examples in the code section, e.g. <a href=\"https://www.kaggle.com/code/jirkaborovec/cancer-subtype-decompose-large-image-tiles\" target=\"_blank\">https://www.kaggle.com/code/jirkaborovec/cancer-subtype-decompose-large-image-tiles</a><br>\nHope this helps.</p>",
      "votes": 3,
      "replies": [
        {
          "id": 2481292,
          "author_name": "Huijie Liu",
          "author_url": "",
          "post_date": "2023-10-14T02:39:18.397000",
          "content": "<p>Thank you. Very good job.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 2478797,
      "author_name": "Seeing Times",
      "author_url": "",
      "post_date": "2023-10-12T08:04:27.743000",
      "content": "<p>Since given GPU in Kaggle is P100 16g which it might not be able to train WSI（Whole Slide Imaging）and even you can it pretty slow I think😬，and you can download data with Kaggle API it‘s may take a night ~</p>",
      "votes": 2,
      "replies": [
        {
          "id": 2479011,
          "author_name": "Huijie Liu",
          "author_url": "",
          "post_date": "2023-10-12T10:41:33.563000",
          "content": "<p>Yeah, so I think local training might be more appropriate.</p>",
          "votes": 1,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2481172": "I did some preprocessing and cropping in a Kaggle notebook and downloaded the resulting patches to do all the training locally. There are already some examples in the code section, e.g. https://www.kaggle.com/code/jirkaborovec/cancer-subtype-decompose-large-image-tiles\nHope this helps.",
    "2478512": "Hello, I would like to ask if there is a way to train locally without downloading data for such big data topics. 🥰",
    "2478797": "Since given GPU in Kaggle is P100 16g which it might not be able to train WSI（Whole Slide Imaging）and even you can it pretty slow I think😬，and you can download data with Kaggle API it‘s may take a night ~"
  }
}