{
  "id": 490930,
  "title": "question on runtime and gpu limit",
  "url": "/competitions/home-credit-credit-risk-model-stability/discussion/490930",
  "author_name": "",
  "post_date": "2024-04-03T23:38:58.444557700Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I have two questions.</p>\n<p>①I believe that until 5/28, my model will be evaluated at 30% of test data. However, it should eventually be evaluated at 100% of the test data. Wouldn't that make the model that I submitted and was able to run within 12 hours invalid because the model eventually needed more time than 12 hours to run?</p>\n<p>②I know that we are limited to 30 hours of gpu usage per week, but even if we use all of this usage time, can we still use the gpu when submitting our models?</p>\n<p>Sorry if these questions have already been answered.</p>",
  "messages": [
    {
      "id": "2734106",
      "postDate": "04/03/2024 23:38:58",
      "content": "<p>I have two questions.</p>\n<p>①I believe that until 5/28, my model will be evaluated at 30% of test data. However, it should eventually be evaluated at 100% of the test data. Wouldn't that make the model that I submitted and was able to run within 12 hours invalid because the model eventually needed more time than 12 hours to run?</p>\n<p>②I know that we are limited to 30 hours of gpu usage per week, but even if we use all of this usage time, can we still use the gpu when submitting our models?</p>\n<p>Sorry if these questions have already been answered.</p>",
      "rawMarkdown": "I have two questions.\n\n①I believe that until 5/28, my model will be evaluated at 30% of test data. However, it should eventually be evaluated at 100% of the test data. Wouldn't that make the model that I submitted and was able to run within 12 hours invalid because the model eventually needed more time than 12 hours to run?\n\n②I know that we are limited to 30 hours of gpu usage per week, but even if we use all of this usage time, can we still use the gpu when submitting our models?\n\nSorry if these questions have already been answered.",
      "votes": null
    },
    {
      "id": "2734607",
      "postDate": "04/04/2024 07:47:08",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/rku0013\" target=\"_blank\">@rku0013</a> ,<br>\nad 1), submissions run on whole dataset, but you see only results of public LB. So \"final\" submissions will not run longer at the end of the competition because of bigger test set</p>",
      "rawMarkdown": "Hi @rku0013 ,\nad 1), submissions run on whole dataset, but you see only results of public LB. So \"final\" submissions will not run longer at the end of the competition because of bigger test set",
      "votes": null
    },
    {
      "id": "2736486",
      "postDate": "04/05/2024 08:12:38",
      "content": "<p>Thank you very much for your detailed information. My questions have been answered.</p>",
      "rawMarkdown": "Thank you very much for your detailed information. My questions have been answered.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2734607,
      "author_name": "tomasjeline2",
      "author_url": "",
      "post_date": "04/04/2024 07:47:08",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/rku0013\" target=\"_blank\">@rku0013</a> ,<br>\nad 1), submissions run on whole dataset, but you see only results of public LB. So \"final\" submissions will not run longer at the end of the competition because of bigger test set</p>",
      "votes": null,
      "replies": [
        {
          "id": 2736486,
          "author_name": "rku0013",
          "author_url": "",
          "post_date": "04/05/2024 08:12:38",
          "content": "<p>Thank you very much for your detailed information. My questions have been answered.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2734106": "I have two questions.\n\n①I believe that until 5/28, my model will be evaluated at 30% of test data. However, it should eventually be evaluated at 100% of the test data. Wouldn't that make the model that I submitted and was able to run within 12 hours invalid because the model eventually needed more time than 12 hours to run?\n\n②I know that we are limited to 30 hours of gpu usage per week, but even if we use all of this usage time, can we still use the gpu when submitting our models?\n\nSorry if these questions have already been answered.",
    "2734607": "Hi @rku0013 ,\nad 1), submissions run on whole dataset, but you see only results of public LB. So \"final\" submissions will not run longer at the end of the competition because of bigger test set",
    "2736486": "Thank you very much for your detailed information. My questions have been answered."
  },
  "source": "meta"
}