{
  "id": 396893,
  "title": "Could I use my computer to generate features for online submission?",
  "url": "/competitions/predict-student-performance-from-game-play/discussion/396893",
  "author_name": "",
  "post_date": "2023-03-23T08:39:08.853500200Z",
  "votes": null,
  "comment_count": 5,
  "views": 0,
  "content": "<p>The memory online is only 8GB, and I often encountered memory errors while doing pre-processing.</p>\n<p>I want to utilize my computer to generate features, and then upload them. And I will write codes in online notebooks with them.</p>\n<p>Does this type of code meet the requirements for the competition?</p>",
  "messages": [
    {
      "id": "2193366",
      "postDate": "03/23/2023 08:39:08",
      "content": "<p>The memory online is only 8GB, and I often encountered memory errors while doing pre-processing.</p>\n<p>I want to utilize my computer to generate features, and then upload them. And I will write codes in online notebooks with them.</p>\n<p>Does this type of code meet the requirements for the competition?</p>",
      "rawMarkdown": "The memory online is only 8GB, and I often encountered memory errors while doing pre-processing.\n\nI want to utilize my computer to generate features, and then upload them. And I will write codes in online notebooks with them.\n\nDoes this type of code meet the requirements for the competition?",
      "votes": null
    },
    {
      "id": "2193396",
      "postDate": "03/23/2023 09:05:45",
      "content": "<p>可以👌👌👌👌👌</p>",
      "rawMarkdown": "可以👌👌👌👌👌",
      "votes": null
    },
    {
      "id": "2193459",
      "postDate": "03/23/2023 10:13:24",
      "content": "<p>Yes. You can also upload trained models files and load models online directly.</p>",
      "rawMarkdown": "Yes. You can also upload trained models files and load models online directly.",
      "votes": null
    },
    {
      "id": "2193817",
      "postDate": "03/23/2023 14:18:39",
      "content": "<p>if i understand you correct then no. You will receive new unseen session_ids while submitting, but to precompute their features you need to have it already downloaded which is impossible. This is not csv competition :) Probably you are getting memory errors doing features for train? There will be no memory issue at submit moment since you will get only one session_id per iteration</p>",
      "rawMarkdown": "if i understand you correct then no. You will receive new unseen session_ids while submitting, but to precompute their features you need to have it already downloaded which is impossible. This is not csv competition :) Probably you are getting memory errors doing features for train? There will be no memory issue at submit moment since you will get only one session_id per iteration",
      "votes": null
    },
    {
      "id": "2198470",
      "postDate": "03/27/2023 04:03:09",
      "content": "<p>Yes, of course, you can train your models on your own device, however, you need to write the inference code on Kaggle as Kaggle will process the code with a hidden test set. </p>",
      "rawMarkdown": "Yes, of course, you can train your models on your own device, however, you need to write the inference code on Kaggle as Kaggle will process the code with a hidden test set.",
      "votes": null
    },
    {
      "id": "2444142",
      "postDate": "09/18/2023 07:28:10",
      "content": "<p>Guess the answer is yes, I am learning,too.</p>",
      "rawMarkdown": "Guess the answer is yes, I am learning,too.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2193396,
      "author_name": "wang5630900",
      "author_url": "",
      "post_date": "03/23/2023 09:05:45",
      "content": "<p>可以👌👌👌👌👌</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2193459,
      "author_name": "hookman",
      "author_url": "",
      "post_date": "03/23/2023 10:13:24",
      "content": "<p>Yes. You can also upload trained models files and load models online directly.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2193817,
      "author_name": "kvlmll",
      "author_url": "",
      "post_date": "03/23/2023 14:18:39",
      "content": "<p>if i understand you correct then no. You will receive new unseen session_ids while submitting, but to precompute their features you need to have it already downloaded which is impossible. This is not csv competition :) Probably you are getting memory errors doing features for train? There will be no memory issue at submit moment since you will get only one session_id per iteration</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2198470,
      "author_name": "harrisonliang",
      "author_url": "",
      "post_date": "03/27/2023 04:03:09",
      "content": "<p>Yes, of course, you can train your models on your own device, however, you need to write the inference code on Kaggle as Kaggle will process the code with a hidden test set. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2444142,
      "author_name": "feliciahuangbebop",
      "author_url": "",
      "post_date": "09/18/2023 07:28:10",
      "content": "<p>Guess the answer is yes, I am learning,too.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2193366": "The memory online is only 8GB, and I often encountered memory errors while doing pre-processing.\n\nI want to utilize my computer to generate features, and then upload them. And I will write codes in online notebooks with them.\n\nDoes this type of code meet the requirements for the competition?",
    "2193396": "可以👌👌👌👌👌",
    "2193459": "Yes. You can also upload trained models files and load models online directly.",
    "2193817": "if i understand you correct then no. You will receive new unseen session_ids while submitting, but to precompute their features you need to have it already downloaded which is impossible. This is not csv competition :) Probably you are getting memory errors doing features for train? There will be no memory issue at submit moment since you will get only one session_id per iteration",
    "2198470": "Yes, of course, you can train your models on your own device, however, you need to write the inference code on Kaggle as Kaggle will process the code with a hidden test set.",
    "2444142": "Guess the answer is yes, I am learning,too."
  },
  "source": "meta"
}