{
  "id": 447958,
  "title": "Private vs. Public LB",
  "url": "/competitions/bengaliai-speech/discussion/447958",
  "author_name": "",
  "post_date": "2023-10-18T00:06:01.134333600Z",
  "votes": 5,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hm, why is private LB score much worse than public LB?</p>",
  "messages": [
    {
      "id": "2486448",
      "postDate": "10/18/2023 00:06:01",
      "content": "<p>Hm, why is private LB score much worse than public LB?</p>",
      "rawMarkdown": "Hm, why is private LB score much worse than public LB?",
      "votes": null
    },
    {
      "id": "2486451",
      "postDate": "10/18/2023 00:08:24",
      "content": "<p>Might be due to differences between the subsets in public vs private set? Some of them might not be in the public at all</p>",
      "rawMarkdown": "Might be due to differences between the subsets in public vs private set? Some of them might not be in the public at all",
      "votes": null
    },
    {
      "id": "2486458",
      "postDate": "10/18/2023 00:12:48",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/tugstugi\" target=\"_blank\">@tugstugi</a> , at first  congratulations! And thanks for putting so much effort into this. Really looking forward to knowing about your approaches in detail. </p>\n<p>About the Private vs Public LB: The private LB split contained samples from some 'challenging' domains more. Samples from all the domains were not present in a similar distribution in both sets.</p>",
      "rawMarkdown": "Hi @tugstugi , at first  congratulations! And thanks for putting so much effort into this. Really looking forward to knowing about your approaches in detail. \n\nAbout the Private vs Public LB: The private LB split contained samples from some 'challenging' domains more. Samples from all the domains were not present in a similar distribution in both sets.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2486451,
      "author_name": "snnclsr",
      "author_url": "",
      "post_date": "10/18/2023 00:08:24",
      "content": "<p>Might be due to differences between the subsets in public vs private set? Some of them might not be in the public at all</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2486458,
      "author_name": "sushmit0109",
      "author_url": "",
      "post_date": "10/18/2023 00:12:48",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/tugstugi\" target=\"_blank\">@tugstugi</a> , at first  congratulations! And thanks for putting so much effort into this. Really looking forward to knowing about your approaches in detail. </p>\n<p>About the Private vs Public LB: The private LB split contained samples from some 'challenging' domains more. Samples from all the domains were not present in a similar distribution in both sets.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2486448": "Hm, why is private LB score much worse than public LB?",
    "2486451": "Might be due to differences between the subsets in public vs private set? Some of them might not be in the public at all",
    "2486458": "Hi @tugstugi , at first  congratulations! And thanks for putting so much effort into this. Really looking forward to knowing about your approaches in detail. \n\nAbout the Private vs Public LB: The private LB split contained samples from some 'challenging' domains more. Samples from all the domains were not present in a similar distribution in both sets."
  },
  "source": "meta"
}