{
  "id": 163820,
  "title": "I cannnot use gpu",
  "url": "/competitions/siim-isic-melanoma-classification/discussion/163820",
  "author_name": "",
  "post_date": "2020-07-03T15:33:43.821360900Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>i use pytorch. \nI set the cuda at the device of model and each data.\nthe script runs without error but it runs on the only cpu.\nI confirm it the draft session and it runs very slowly.\nwhy and how can i use it</p>",
  "messages": [
    {
      "id": "914041",
      "postDate": "07/03/2020 15:33:43",
      "content": "<p>i use pytorch. \nI set the cuda at the device of model and each data.\nthe script runs without error but it runs on the only cpu.\nI confirm it the draft session and it runs very slowly.\nwhy and how can i use it</p>",
      "rawMarkdown": "i use pytorch. \nI set the cuda at the device of model and each data.\nthe script runs without error but it runs on the only cpu.\nI confirm it the draft session and it runs very slowly.\nwhy and how can i use it",
      "votes": null
    },
    {
      "id": "914064",
      "postDate": "07/03/2020 15:52:42",
      "content": "<p>What data are you using, if you are using the input data from the competition directly, it must be taking a lot of time in loading and resizing. </p>\n\n<p>Use any pre-processed data otherwise, it would be good to go then!</p>",
      "rawMarkdown": "What data are you using, if you are using the input data from the competition directly, it must be taking a lot of time in loading and resizing. \n\nUse any pre-processed data otherwise, it would be good to go then!",
      "votes": null
    },
    {
      "id": "914087",
      "postDate": "07/03/2020 16:08:38",
      "content": "<p>Hey <a href=\"/sorannakagawa\">@sorannakagawa</a>, it is a big hard to help you debug without code snippet.\nWhen all your data/models are loaded and transfered to your gpu, can you try running: !nvidia-smi in a cell and post here the output ?</p>",
      "rawMarkdown": "Hey @sorannakagawa, it is a big hard to help you debug without code snippet.\nWhen all your data/models are loaded and transfered to your gpu, can you try running: !nvidia-smi in a cell and post here the output ?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 914064,
      "author_name": "sarques",
      "author_url": "",
      "post_date": "07/03/2020 15:52:42",
      "content": "<p>What data are you using, if you are using the input data from the competition directly, it must be taking a lot of time in loading and resizing. </p>\n\n<p>Use any pre-processed data otherwise, it would be good to go then!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 914087,
      "author_name": "dimartinot",
      "author_url": "",
      "post_date": "07/03/2020 16:08:38",
      "content": "<p>Hey <a href=\"/sorannakagawa\">@sorannakagawa</a>, it is a big hard to help you debug without code snippet.\nWhen all your data/models are loaded and transfered to your gpu, can you try running: !nvidia-smi in a cell and post here the output ?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "914041": "i use pytorch. \nI set the cuda at the device of model and each data.\nthe script runs without error but it runs on the only cpu.\nI confirm it the draft session and it runs very slowly.\nwhy and how can i use it",
    "914064": "What data are you using, if you are using the input data from the competition directly, it must be taking a lot of time in loading and resizing. \n\nUse any pre-processed data otherwise, it would be good to go then!",
    "914087": "Hey @sorannakagawa, it is a big hard to help you debug without code snippet.\nWhen all your data/models are loaded and transfered to your gpu, can you try running: !nvidia-smi in a cell and post here the output ?"
  },
  "source": "meta"
}