{
  "id": 95442,
  "title": "How to use GPU to accelerate the training?",
  "url": "/competitions/champs-scalar-coupling/discussion/95442",
  "author_name": "",
  "post_date": "2019-06-12T09:28:39.501523500Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>When I turn the GPU on , it tells me 'Custom packages are not supported for GPU instances.'....\nBut why ?  </p>",
  "messages": [
    {
      "id": "551044",
      "postDate": "06/12/2019 09:28:39",
      "content": "<p>When I turn the GPU on , it tells me 'Custom packages are not supported for GPU instances.'....\nBut why ?  </p>",
      "rawMarkdown": "When I turn the GPU on , it tells me 'Custom packages are not supported for GPU instances.'....\nBut why ?",
      "votes": null
    },
    {
      "id": "551159",
      "postDate": "06/12/2019 12:05:52",
      "content": "<p>I could not find a explanation for \"why\" but some said in <a href=\"https://github.com/Kaggle/docker-python/issues/508\">https://github.com/Kaggle/docker-python/issues/508</a>\nthat you can do it turning Internet access and using <code>!pip install my-package</code>, i. e., you are not forbidden but just can not do it by the user interface.</p>",
      "rawMarkdown": "I could not find a explanation for \"why\" but some said in https://github.com/Kaggle/docker-python/issues/508\nthat you can do it turning Internet access and using `!pip install my-package`, i. e., you are not forbidden but just can not do it by the user interface.",
      "votes": null
    },
    {
      "id": "551653",
      "postDate": "06/13/2019 01:33:50",
      "content": "<p>Thanks a lot , I have solved this problem already!</p>",
      "rawMarkdown": "Thanks a lot , I have solved this problem already!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 551159,
      "author_name": "alexpcli",
      "author_url": "",
      "post_date": "06/12/2019 12:05:52",
      "content": "<p>I could not find a explanation for \"why\" but some said in <a href=\"https://github.com/Kaggle/docker-python/issues/508\">https://github.com/Kaggle/docker-python/issues/508</a>\nthat you can do it turning Internet access and using <code>!pip install my-package</code>, i. e., you are not forbidden but just can not do it by the user interface.</p>",
      "votes": null,
      "replies": [
        {
          "id": 551653,
          "author_name": "luyouyou",
          "author_url": "",
          "post_date": "06/13/2019 01:33:50",
          "content": "<p>Thanks a lot , I have solved this problem already!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "551044": "When I turn the GPU on , it tells me 'Custom packages are not supported for GPU instances.'....\nBut why ?",
    "551159": "I could not find a explanation for \"why\" but some said in https://github.com/Kaggle/docker-python/issues/508\nthat you can do it turning Internet access and using `!pip install my-package`, i. e., you are not forbidden but just can not do it by the user interface.",
    "551653": "Thanks a lot , I have solved this problem already!"
  },
  "source": "meta"
}