{
  "id": 452540,
  "title": "K-fold predictions",
  "url": "/competitions/predict-ai-model-runtime/discussion/452540",
  "author_name": "",
  "post_date": "2023-11-02T13:58:48.107374700Z",
  "votes": 8,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Did you guys find K-fold helpful for the layout predictions? I tried it and against my intuition found no improvement in my results.<br>\nWould love to hear your thoughts on it.</p>",
  "messages": [
    {
      "id": "2509670",
      "postDate": "11/02/2023 13:58:48",
      "content": "<p>Did you guys find K-fold helpful for the layout predictions? I tried it and against my intuition found no improvement in my results.<br>\nWould love to hear your thoughts on it.</p>",
      "rawMarkdown": "Did you guys find K-fold helpful for the layout predictions? I tried it and against my intuition found no improvement in my results.\nWould love to hear your thoughts on it.",
      "votes": null
    },
    {
      "id": "2516610",
      "postDate": "11/07/2023 20:14:05",
      "content": "<p>We observed no improvement from K-fold on the layout with both runtime averaging and MRR aggregation. Not to say, it is very expensive to train.</p>",
      "rawMarkdown": "We observed no improvement from K-fold on the layout with both runtime averaging and MRR aggregation. Not to say, it is very expensive to train.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2516610,
      "author_name": "dmitriikhizbullin",
      "author_url": "",
      "post_date": "11/07/2023 20:14:05",
      "content": "<p>We observed no improvement from K-fold on the layout with both runtime averaging and MRR aggregation. Not to say, it is very expensive to train.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2509670": "Did you guys find K-fold helpful for the layout predictions? I tried it and against my intuition found no improvement in my results.\nWould love to hear your thoughts on it.",
    "2516610": "We observed no improvement from K-fold on the layout with both runtime averaging and MRR aggregation. Not to say, it is very expensive to train."
  },
  "source": "meta"
}