{
  "id": 175907,
  "title": "Single fold + single scale better than multi-fold + multi scale?",
  "url": "/competitions/global-wheat-detection/discussion/175907",
  "author_name": "",
  "post_date": "2020-08-19T19:46:18.991799700Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I managed to do a late submission before the host disabled it 2 weeks ago. It was a single fold + single-scale inference (1.25x) without pseudo-label. It gave me 0.6774 on private.<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2F872c220ef90c38a15e628a662322ded9%2Fsingle.png?generation=1597866326880832&amp;alt=media\" alt=\"\"></p>\n<p>The selected one was 3 folds + multi-scale (1.1x + 1.25x + 1.35x) without pseudo-label. It gave me 0.6504. I guess better luck next time<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2F195bc752dace9abd76dd362fcdc18cb4%2Fmulti.png?generation=1597866375225891&amp;alt=media\" alt=\"\"></p>\n<p>Update: just took a look at my 5-fold ensemble…it is 0.6956</p>",
  "messages": [
    {
      "id": "977918",
      "postDate": "08/19/2020 19:46:18",
      "content": "<p>I managed to do a late submission before the host disabled it 2 weeks ago. It was a single fold + single-scale inference (1.25x) without pseudo-label. It gave me 0.6774 on private.<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2F872c220ef90c38a15e628a662322ded9%2Fsingle.png?generation=1597866326880832&amp;alt=media\" alt=\"\"></p>\n<p>The selected one was 3 folds + multi-scale (1.1x + 1.25x + 1.35x) without pseudo-label. It gave me 0.6504. I guess better luck next time<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2F195bc752dace9abd76dd362fcdc18cb4%2Fmulti.png?generation=1597866375225891&amp;alt=media\" alt=\"\"></p>\n<p>Update: just took a look at my 5-fold ensemble…it is 0.6956</p>",
      "rawMarkdown": "I managed to do a late submission before the host disabled it 2 weeks ago. It was a single fold + single-scale inference (1.25x) without pseudo-label. It gave me 0.6774 on private.\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2F872c220ef90c38a15e628a662322ded9%2Fsingle.png?generation=1597866326880832&alt=media)\n\nThe selected one was 3 folds + multi-scale (1.1x + 1.25x + 1.35x) without pseudo-label. It gave me 0.6504. I guess better luck next time\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2F195bc752dace9abd76dd362fcdc18cb4%2Fmulti.png?generation=1597866375225891&alt=media)\n\nUpdate: just took a look at my 5-fold ensemble...it is 0.6956",
      "votes": null
    },
    {
      "id": "977941",
      "postDate": "08/19/2020 20:05:34",
      "content": "<p>And pseudo-label worked perfectly on private… My heart's bleeding 😭</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2Faead6fe662aa331a628365141386ce4f%2Fpseudo.png?generation=1597867485226613&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "And pseudo-label worked perfectly on private... My heart's bleeding 😭\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2Faead6fe662aa331a628365141386ce4f%2Fpseudo.png?generation=1597867485226613&alt=media)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 977941,
      "author_name": "jihangz",
      "author_url": "",
      "post_date": "08/19/2020 20:05:34",
      "content": "<p>And pseudo-label worked perfectly on private… My heart's bleeding 😭</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2Faead6fe662aa331a628365141386ce4f%2Fpseudo.png?generation=1597867485226613&amp;alt=media\" alt=\"\"></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "977918": "I managed to do a late submission before the host disabled it 2 weeks ago. It was a single fold + single-scale inference (1.25x) without pseudo-label. It gave me 0.6774 on private.\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2F872c220ef90c38a15e628a662322ded9%2Fsingle.png?generation=1597866326880832&alt=media)\n\nThe selected one was 3 folds + multi-scale (1.1x + 1.25x + 1.35x) without pseudo-label. It gave me 0.6504. I guess better luck next time\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2F195bc752dace9abd76dd362fcdc18cb4%2Fmulti.png?generation=1597866375225891&alt=media)\n\nUpdate: just took a look at my 5-fold ensemble...it is 0.6956",
    "977941": "And pseudo-label worked perfectly on private... My heart's bleeding 😭\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1605499%2Faead6fe662aa331a628365141386ce4f%2Fpseudo.png?generation=1597867485226613&alt=media)"
  },
  "source": "meta"
}