{
  "id": 151400,
  "title": "Limit about CPU, GPU",
  "url": "/competitions/global-wheat-detection/discussion/151400",
  "author_name": "",
  "post_date": "2020-05-15T09:59:42.704391300Z",
  "votes": 1,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hi,\nSorry I am newbie,</p>\n\n<p>Is limiting the amount of time for Compute only apply for the inference's notebook ?\n or both (and the train's notebook) ?</p>\n\n<p>Thanks.</p>",
  "messages": [
    {
      "id": "848873",
      "postDate": "05/15/2020 09:59:42",
      "content": "<p>Hi,\nSorry I am newbie,</p>\n\n<p>Is limiting the amount of time for Compute only apply for the inference's notebook ?\n or both (and the train's notebook) ?</p>\n\n<p>Thanks.</p>",
      "rawMarkdown": "Hi,\nSorry I am newbie,\n\nIs limiting the amount of time for Compute only apply for the inference's notebook ?\n or both (and the train's notebook) ?\n\nThanks.",
      "votes": null
    },
    {
      "id": "848920",
      "postDate": "05/15/2020 10:54:04",
      "content": "<p>only inference</p>",
      "rawMarkdown": "only inference",
      "votes": null
    },
    {
      "id": "849165",
      "postDate": "05/15/2020 14:39:39",
      "content": "<p>I thought pre-trained models/external data have to be freely &amp; publicly available or am I misunderstanding the code requirements?</p>",
      "rawMarkdown": "I thought pre-trained models/external data have to be freely &amp; publicly available or am I misunderstanding the code requirements?",
      "votes": null
    },
    {
      "id": "849172",
      "postDate": "05/15/2020 14:47:24",
      "content": "<p>you are correct. we have to post our pretrained-model/external data in <a href=\"https://www.kaggle.com/c/global-wheat-detection/discussion/148561\">this thread</a></p>",
      "rawMarkdown": "you are correct. we have to post our pretrained-model/external data in [this thread](https://www.kaggle.com/c/global-wheat-detection/discussion/148561)",
      "votes": null
    },
    {
      "id": "849217",
      "postDate": "05/15/2020 15:29:40",
      "content": "<p>Are you allowed to train in one kernel and then inference in another? I seem to be seeing that a lot in the shared notebooks. Most people are doing that to escape the internet restrictions, but what if the training notebook is private. Is that allowed?</p>",
      "rawMarkdown": "Are you allowed to train in one kernel and then inference in another? I seem to be seeing that a lot in the shared notebooks. Most people are doing that to escape the internet restrictions, but what if the training notebook is private. Is that allowed?",
      "votes": null
    },
    {
      "id": "849241",
      "postDate": "05/15/2020 15:42:42",
      "content": "<p>absolutely, you can keep both training and inference notebook privately. </p>",
      "rawMarkdown": "absolutely, you can keep both training and inference notebook privately.",
      "votes": null
    },
    {
      "id": "849851",
      "postDate": "05/16/2020 05:52:04",
      "content": "<p>I never compete in this type of competition, but it is allowed to create the model on one notebook (without restrictions) and to load it on another inference notebook, that has this GPU &amp; CPU restrictions? (what is the advantage by using 2 separate notebooks, that everyone do?)</p>\n\n<p>another question, why there are restrictions both GPU &amp;CPU?</p>",
      "rawMarkdown": "I never compete in this type of competition, but it is allowed to create the model on one notebook (without restrictions) and to load it on another inference notebook, that has this GPU &amp; CPU restrictions? (what is the advantage by using 2 separate notebooks, that everyone do?)\n\nanother question, why there are restrictions both GPU &amp;CPU?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 848920,
      "author_name": "nvnnghia",
      "author_url": "",
      "post_date": "05/15/2020 10:54:04",
      "content": "<p>only inference</p>",
      "votes": null,
      "replies": [
        {
          "id": 849165,
          "author_name": "tommy260",
          "author_url": "",
          "post_date": "05/15/2020 14:39:39",
          "content": "<p>I thought pre-trained models/external data have to be freely &amp; publicly available or am I misunderstanding the code requirements?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 849172,
          "author_name": "nvnnghia",
          "author_url": "",
          "post_date": "05/15/2020 14:47:24",
          "content": "<p>you are correct. we have to post our pretrained-model/external data in <a href=\"https://www.kaggle.com/c/global-wheat-detection/discussion/148561\">this thread</a></p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 849217,
          "author_name": "tommy260",
          "author_url": "",
          "post_date": "05/15/2020 15:29:40",
          "content": "<p>Are you allowed to train in one kernel and then inference in another? I seem to be seeing that a lot in the shared notebooks. Most people are doing that to escape the internet restrictions, but what if the training notebook is private. Is that allowed?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 849241,
          "author_name": "nvnnghia",
          "author_url": "",
          "post_date": "05/15/2020 15:42:42",
          "content": "<p>absolutely, you can keep both training and inference notebook privately. </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 849851,
      "author_name": "tomeramit2",
      "author_url": "",
      "post_date": "05/16/2020 05:52:04",
      "content": "<p>I never compete in this type of competition, but it is allowed to create the model on one notebook (without restrictions) and to load it on another inference notebook, that has this GPU &amp; CPU restrictions? (what is the advantage by using 2 separate notebooks, that everyone do?)</p>\n\n<p>another question, why there are restrictions both GPU &amp;CPU?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "848873": "Hi,\nSorry I am newbie,\n\nIs limiting the amount of time for Compute only apply for the inference's notebook ?\n or both (and the train's notebook) ?\n\nThanks.",
    "848920": "only inference",
    "849165": "I thought pre-trained models/external data have to be freely &amp; publicly available or am I misunderstanding the code requirements?",
    "849172": "you are correct. we have to post our pretrained-model/external data in [this thread](https://www.kaggle.com/c/global-wheat-detection/discussion/148561)",
    "849217": "Are you allowed to train in one kernel and then inference in another? I seem to be seeing that a lot in the shared notebooks. Most people are doing that to escape the internet restrictions, but what if the training notebook is private. Is that allowed?",
    "849241": "absolutely, you can keep both training and inference notebook privately.",
    "849851": "I never compete in this type of competition, but it is allowed to create the model on one notebook (without restrictions) and to load it on another inference notebook, that has this GPU &amp; CPU restrictions? (what is the advantage by using 2 separate notebooks, that everyone do?)\n\nanother question, why there are restrictions both GPU &amp;CPU?"
  },
  "source": "meta"
}