{
  "id": 498506,
  "title": "What platform you train your model?",
  "url": "/competitions/birdclef-2024/discussion/498506",
  "author_name": "",
  "post_date": "2024-04-28T15:11:42.450517Z",
  "votes": 3,
  "comment_count": 3,
  "views": 0,
  "content": "<p>The competition data is big and Kaggle has time limitation each week. What platform you would train your model efficiently?</p>",
  "messages": [
    {
      "id": "2781111",
      "postDate": "04/28/2024 15:11:42",
      "content": "<p>The competition data is big and Kaggle has time limitation each week. What platform you would train your model efficiently?</p>",
      "rawMarkdown": "The competition data is big and Kaggle has time limitation each week. What platform you would train your model efficiently?",
      "votes": null
    },
    {
      "id": "2781456",
      "postDate": "04/28/2024 19:44:00",
      "content": "<p>Cloud services are handy in such times. I use my local PC too and Colab, runpod.io, vast.ai, etc. for training models based on my requirements as on date <a href=\"https://www.kaggle.com/yiding111\" target=\"_blank\">@yiding111</a> <br>\nI believe I just saw this question on another forum too!</p>",
      "rawMarkdown": "Cloud services are handy in such times. I use my local PC too and Colab, runpod.io, vast.ai, etc. for training models based on my requirements as on date @yiding111 \nI believe I just saw this question on another forum too!",
      "votes": null
    },
    {
      "id": "2781646",
      "postDate": "04/28/2024 22:51:21",
      "content": "<p>I've used kaggle notebooks exclusively. It helps me keep a clear record of my work, and overall I&nbsp;prefer not dealing with a local setup. So far I haven't come close to hitting GPU limits, but maybe I'll get near to that later in the competition. </p>\n<p>Another issue is that Kaggle machines have more limited disk space than I would like &amp; sometimes unpredictable disk read times, so optimizing dataloading/precomputing probably requires more thought than with other setups.</p>",
      "rawMarkdown": "I've used kaggle notebooks exclusively. It helps me keep a clear record of my work, and overall I prefer not dealing with a local setup. So far I haven't come close to hitting GPU limits, but maybe I'll get near to that later in the competition. \n\nAnother issue is that Kaggle machines have more limited disk space than I would like & sometimes unpredictable disk read times, so optimizing dataloading/precomputing probably requires more thought than with other setups.",
      "votes": null
    },
    {
      "id": "2781856",
      "postDate": "04/29/2024 03:22:17",
      "content": "<p>Yes, that my bad. Sorry for that. I wanted to post it on Q&amp;A forum, but in the wrong place.</p>",
      "rawMarkdown": "Yes, that my bad. Sorry for that. I wanted to post it on Q&A forum, but in the wrong place.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2781456,
      "author_name": "ravi20076",
      "author_url": "",
      "post_date": "04/28/2024 19:44:00",
      "content": "<p>Cloud services are handy in such times. I use my local PC too and Colab, runpod.io, vast.ai, etc. for training models based on my requirements as on date <a href=\"https://www.kaggle.com/yiding111\" target=\"_blank\">@yiding111</a> <br>\nI believe I just saw this question on another forum too!</p>",
      "votes": null,
      "replies": [
        {
          "id": 2781856,
          "author_name": "yiding111",
          "author_url": "",
          "post_date": "04/29/2024 03:22:17",
          "content": "<p>Yes, that my bad. Sorry for that. I wanted to post it on Q&amp;A forum, but in the wrong place.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 2781646,
      "author_name": "robbynevels",
      "author_url": "",
      "post_date": "04/28/2024 22:51:21",
      "content": "<p>I've used kaggle notebooks exclusively. It helps me keep a clear record of my work, and overall I&nbsp;prefer not dealing with a local setup. So far I haven't come close to hitting GPU limits, but maybe I'll get near to that later in the competition. </p>\n<p>Another issue is that Kaggle machines have more limited disk space than I would like &amp; sometimes unpredictable disk read times, so optimizing dataloading/precomputing probably requires more thought than with other setups.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2781111": "The competition data is big and Kaggle has time limitation each week. What platform you would train your model efficiently?",
    "2781456": "Cloud services are handy in such times. I use my local PC too and Colab, runpod.io, vast.ai, etc. for training models based on my requirements as on date @yiding111 \nI believe I just saw this question on another forum too!",
    "2781646": "I've used kaggle notebooks exclusively. It helps me keep a clear record of my work, and overall I prefer not dealing with a local setup. So far I haven't come close to hitting GPU limits, but maybe I'll get near to that later in the competition. \n\nAnother issue is that Kaggle machines have more limited disk space than I would like & sometimes unpredictable disk read times, so optimizing dataloading/precomputing probably requires more thought than with other setups.",
    "2781856": "Yes, that my bad. Sorry for that. I wanted to post it on Q&A forum, but in the wrong place."
  },
  "source": "meta"
}