{
  "id": 198005,
  "title": "This competition limits how much memory ？",
  "url": "/competitions/riiid-test-answer-prediction/discussion/198005",
  "author_name": "",
  "post_date": "2020-11-19T08:40:20.657421200Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>1.How much memory does this competition limit the program to use?</p>\n<p>2.Must it be run on Kaggle's notebook? </p>\n<p>Why do I use 2G of memory (calculated by the <strong>sizeof</strong>() function), and it shows that I use 8G of memory on Kaggle's notebook? This makes my program very easy to exceed the 16G memory limit, I am very confused.Can someone help me solve this problem?</p>\n<p>Thanks!</p>",
  "messages": [
    {
      "id": "1083688",
      "postDate": "11/19/2020 08:40:20",
      "content": "<p>1.How much memory does this competition limit the program to use?</p>\n<p>2.Must it be run on Kaggle's notebook? </p>\n<p>Why do I use 2G of memory (calculated by the <strong>sizeof</strong>() function), and it shows that I use 8G of memory on Kaggle's notebook? This makes my program very easy to exceed the 16G memory limit, I am very confused.Can someone help me solve this problem?</p>\n<p>Thanks!</p>",
      "rawMarkdown": "1.How much memory does this competition limit the program to use?\n\n2.Must it be run on Kaggle's notebook? \n\nWhy do I use 2G of memory (calculated by the __sizeof__() function), and it shows that I use 8G of memory on Kaggle's notebook? This makes my program very easy to exceed the 16G memory limit, I am very confused.Can someone help me solve this problem?\n\nThanks!",
      "votes": null
    },
    {
      "id": "1083690",
      "postDate": "11/19/2020 08:42:14",
      "content": "<p>Hope to get a reply soon！</p>",
      "rawMarkdown": "Hope to get a reply soon！",
      "votes": null
    },
    {
      "id": "1083794",
      "postDate": "11/19/2020 11:47:09",
      "content": "<p>It is better to answer the 2nd question first.</p>\n<p>You DON'T have to train on Kaggle's notebook, but you MUST make predictions on Kaggle notebook (sometimes we call it Inference), and its runtime must not exceed the requirements. You can train on your local machine, or anywhere, and save the pretrained model to a dataset. You make a notebook that uses that dataset, and make predictions to submit.</p>\n<p>The first question's limit is the Kaggle system itself. Kaggle notebook has limited Memory, so to tackle that problem is what we must do it ourselves. </p>",
      "rawMarkdown": "It is better to answer the 2nd question first.\n\nYou DON'T have to train on Kaggle's notebook, but you MUST make predictions on Kaggle notebook (sometimes we call it Inference), and its runtime must not exceed the requirements. You can train on your local machine, or anywhere, and save the pretrained model to a dataset. You make a notebook that uses that dataset, and make predictions to submit.\n\nThe first question's limit is the Kaggle system itself. Kaggle notebook has limited Memory, so to tackle that problem is what we must do it ourselves.",
      "votes": null
    },
    {
      "id": "1083818",
      "postDate": "11/19/2020 12:21:32",
      "content": "<p>Thanks !<br>\nI just find that the Kaggle’s notebook sometimes displays the system information delayed.</p>",
      "rawMarkdown": "Thanks !\nI just find that the Kaggle’s notebook sometimes displays the system information delayed.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1083690,
      "author_name": "berlinli",
      "author_url": "",
      "post_date": "11/19/2020 08:42:14",
      "content": "<p>Hope to get a reply soon！</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1083794,
      "author_name": "aeryss",
      "author_url": "",
      "post_date": "11/19/2020 11:47:09",
      "content": "<p>It is better to answer the 2nd question first.</p>\n<p>You DON'T have to train on Kaggle's notebook, but you MUST make predictions on Kaggle notebook (sometimes we call it Inference), and its runtime must not exceed the requirements. You can train on your local machine, or anywhere, and save the pretrained model to a dataset. You make a notebook that uses that dataset, and make predictions to submit.</p>\n<p>The first question's limit is the Kaggle system itself. Kaggle notebook has limited Memory, so to tackle that problem is what we must do it ourselves. </p>",
      "votes": null,
      "replies": [
        {
          "id": 1083818,
          "author_name": "berlinli",
          "author_url": "",
          "post_date": "11/19/2020 12:21:32",
          "content": "<p>Thanks !<br>\nI just find that the Kaggle’s notebook sometimes displays the system information delayed.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1083688": "1.How much memory does this competition limit the program to use?\n\n2.Must it be run on Kaggle's notebook? \n\nWhy do I use 2G of memory (calculated by the __sizeof__() function), and it shows that I use 8G of memory on Kaggle's notebook? This makes my program very easy to exceed the 16G memory limit, I am very confused.Can someone help me solve this problem?\n\nThanks!",
    "1083690": "Hope to get a reply soon！",
    "1083794": "It is better to answer the 2nd question first.\n\nYou DON'T have to train on Kaggle's notebook, but you MUST make predictions on Kaggle notebook (sometimes we call it Inference), and its runtime must not exceed the requirements. You can train on your local machine, or anywhere, and save the pretrained model to a dataset. You make a notebook that uses that dataset, and make predictions to submit.\n\nThe first question's limit is the Kaggle system itself. Kaggle notebook has limited Memory, so to tackle that problem is what we must do it ourselves.",
    "1083818": "Thanks !\nI just find that the Kaggle’s notebook sometimes displays the system information delayed."
  },
  "source": "meta"
}