{
  "id": 272516,
  "title": "Can I use my local machine to train a model?",
  "url": "/competitions/landmark-retrieval-2021/discussion/272516",
  "author_name": "",
  "post_date": "2021-09-15T19:50:07.278138900Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I have an access to brand new gaming computer which has got Nvidia GTX  3050 Ti GPU. As I am realitively new to kaggle, after reading <strong>coding requirements</strong> which puts some restrictions about pretrained models, I am wondering how I can use this machine to train a model for hours to get higher score. </p>\n<p>According to my thinking, I will download all the data, build a model, train it for hours/days and after optimizing I will upload the saved model to generate a submission file.</p>\n<p>In this case I have questions: </p>\n<ol>\n<li>Should I train a model in my machine upto 12 hour as told in code requirments or unlimited ?</li>\n<li>If I just upload my pre-trained model, isn't it considered as a private data?</li>\n<li>or should I also publicly provide it on kaggle datasets?</li>\n</ol>\n<p>if my hyposesis is wrong, is there any task I can use my machine for getting higher ranks in a leaderboard?</p>\n<p>Thank you very much for replying!<br>\nGood luck to everyone!</p>",
  "messages": [
    {
      "id": "1514245",
      "postDate": "09/15/2021 19:50:07",
      "content": "<p>I have an access to brand new gaming computer which has got Nvidia GTX  3050 Ti GPU. As I am realitively new to kaggle, after reading <strong>coding requirements</strong> which puts some restrictions about pretrained models, I am wondering how I can use this machine to train a model for hours to get higher score. </p>\n<p>According to my thinking, I will download all the data, build a model, train it for hours/days and after optimizing I will upload the saved model to generate a submission file.</p>\n<p>In this case I have questions: </p>\n<ol>\n<li>Should I train a model in my machine upto 12 hour as told in code requirments or unlimited ?</li>\n<li>If I just upload my pre-trained model, isn't it considered as a private data?</li>\n<li>or should I also publicly provide it on kaggle datasets?</li>\n</ol>\n<p>if my hyposesis is wrong, is there any task I can use my machine for getting higher ranks in a leaderboard?</p>\n<p>Thank you very much for replying!<br>\nGood luck to everyone!</p>",
      "rawMarkdown": "I have an access to brand new gaming computer which has got Nvidia GTX  3050 Ti GPU. As I am realitively new to kaggle, after reading **coding requirements** which puts some restrictions about pretrained models, I am wondering how I can use this machine to train a model for hours to get higher score. \n\nAccording to my thinking, I will download all the data, build a model, train it for hours/days and after optimizing I will upload the saved model to generate a submission file.\n\nIn this case I have questions: \n1. Should I train a model in my machine upto 12 hour as told in code requirments or unlimited ?\n2. If I just upload my pre-trained model, isn't it considered as a private data?\n3. or should I also publicly provide it on kaggle datasets?\n\nif my hyposesis is wrong, is there any task I can use my machine for getting higher ranks in a leaderboard?\n\n\n\nThank you very much for replying!\nGood luck to everyone!",
      "votes": null
    },
    {
      "id": "1514328",
      "postDate": "09/16/2021 01:15:40",
      "content": "<p>You are free to train your model for as long as you want. The code requirements are for inference time.<br>\nFor this competition, you need to submit an inference notebook and your inference notebook should finish running within the time specified in the code requirements.<br>\nYou can refer to <a href=\"https://www.kaggle.com/hidehisaarai1213/glret21-efficientnetb0-baseline-inference\" target=\"_blank\">this kernel</a></p>\n<p>Also, you don't need to publish your trained models anywhere<br>\nGood luck!</p>",
      "rawMarkdown": "You are free to train your model for as long as you want. The code requirements are for inference time.\nFor this competition, you need to submit an inference notebook and your inference notebook should finish running within the time specified in the code requirements.\nYou can refer to [this kernel](https://www.kaggle.com/hidehisaarai1213/glret21-efficientnetb0-baseline-inference)\n\nAlso, you don't need to publish your trained models anywhere\nGood luck!",
      "votes": null
    },
    {
      "id": "1514385",
      "postDate": "09/16/2021 03:20:28",
      "content": "<p>Thank you bro for the reply. another question does inference take alot time. Because I think as long as we do not get gradient it takes much more less time even if we train for several days.</p>",
      "rawMarkdown": "Thank you bro for the reply. another question does inference take alot time. Because I think as long as we do not get gradient it takes much more less time even if we train for several days.",
      "votes": null
    },
    {
      "id": "1514404",
      "postDate": "09/16/2021 03:51:07",
      "content": "<p>Yes you are right. Inference doesn't take much time but the time constraints are there in case you plan on using ensembles of various models.</p>",
      "rawMarkdown": "Yes you are right. Inference doesn't take much time but the time constraints are there in case you plan on using ensembles of various models.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1514328,
      "author_name": "debarshichanda",
      "author_url": "",
      "post_date": "09/16/2021 01:15:40",
      "content": "<p>You are free to train your model for as long as you want. The code requirements are for inference time.<br>\nFor this competition, you need to submit an inference notebook and your inference notebook should finish running within the time specified in the code requirements.<br>\nYou can refer to <a href=\"https://www.kaggle.com/hidehisaarai1213/glret21-efficientnetb0-baseline-inference\" target=\"_blank\">this kernel</a></p>\n<p>Also, you don't need to publish your trained models anywhere<br>\nGood luck!</p>",
      "votes": null,
      "replies": [
        {
          "id": 1514385,
          "author_name": "abbosjon",
          "author_url": "",
          "post_date": "09/16/2021 03:20:28",
          "content": "<p>Thank you bro for the reply. another question does inference take alot time. Because I think as long as we do not get gradient it takes much more less time even if we train for several days.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1514404,
          "author_name": "debarshichanda",
          "author_url": "",
          "post_date": "09/16/2021 03:51:07",
          "content": "<p>Yes you are right. Inference doesn't take much time but the time constraints are there in case you plan on using ensembles of various models.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1514245": "I have an access to brand new gaming computer which has got Nvidia GTX  3050 Ti GPU. As I am realitively new to kaggle, after reading **coding requirements** which puts some restrictions about pretrained models, I am wondering how I can use this machine to train a model for hours to get higher score. \n\nAccording to my thinking, I will download all the data, build a model, train it for hours/days and after optimizing I will upload the saved model to generate a submission file.\n\nIn this case I have questions: \n1. Should I train a model in my machine upto 12 hour as told in code requirments or unlimited ?\n2. If I just upload my pre-trained model, isn't it considered as a private data?\n3. or should I also publicly provide it on kaggle datasets?\n\nif my hyposesis is wrong, is there any task I can use my machine for getting higher ranks in a leaderboard?\n\n\n\nThank you very much for replying!\nGood luck to everyone!",
    "1514328": "You are free to train your model for as long as you want. The code requirements are for inference time.\nFor this competition, you need to submit an inference notebook and your inference notebook should finish running within the time specified in the code requirements.\nYou can refer to [this kernel](https://www.kaggle.com/hidehisaarai1213/glret21-efficientnetb0-baseline-inference)\n\nAlso, you don't need to publish your trained models anywhere\nGood luck!",
    "1514385": "Thank you bro for the reply. another question does inference take alot time. Because I think as long as we do not get gradient it takes much more less time even if we train for several days.",
    "1514404": "Yes you are right. Inference doesn't take much time but the time constraints are there in case you plan on using ensembles of various models."
  },
  "source": "meta"
}