{
  "id": 240622,
  "title": "EfficientNet B4 MixUp CV 0.98 LB 0.97",
  "url": "/competitions/seti-breakthrough-listen/discussion/240622",
  "author_name": "Salman",
  "post_date": "2021-05-20T16:52:28.427000",
  "votes": 9,
  "comment_count": 2,
  "views": 0,
  "content": "<ul>\n<li><p>I have shared EfficientNet B4 trained with MixUP for 10 Epochs achieving 0.97 LB.</p></li>\n<li><p>I shared training function, way to use MixUp in <a href=\"https://www.kaggle.com/micheomaano/efficientnet-b4-mixup-cv-0-98-lb-0-97\" target=\"_blank\">Notebook</a></p></li>\n<li><p>This notebook can be used for experiments on local machines to train a 5 fold to achieve 0.98.</p></li>\n</ul>\n<p>I am working on signal processing to mix some features with images. I will share update on that as well.</p>\n<p>Happy Coding.</p>",
  "messages": [
    {
      "id": 1316565,
      "postDate": "2021-05-20T16:52:28.427Z",
      "content": "<ul>\n<li><p>I have shared EfficientNet B4 trained with MixUP for 10 Epochs achieving 0.97 LB.</p></li>\n<li><p>I shared training function, way to use MixUp in <a href=\"https://www.kaggle.com/micheomaano/efficientnet-b4-mixup-cv-0-98-lb-0-97\" target=\"_blank\">Notebook</a></p></li>\n<li><p>This notebook can be used for experiments on local machines to train a 5 fold to achieve 0.98.</p></li>\n</ul>\n<p>I am working on signal processing to mix some features with images. I will share update on that as well.</p>\n<p>Happy Coding.</p>",
      "rawMarkdown": "- I have shared EfficientNet B4 trained with MixUP for 10 Epochs achieving 0.97 LB.\n\n- I shared training function, way to use MixUp in [Notebook](https://www.kaggle.com/micheomaano/efficientnet-b4-mixup-cv-0-98-lb-0-97)\n\n- This notebook can be used for experiments on local machines to train a 5 fold to achieve 0.98.\n\nI am working on signal processing to mix some features with images. I will share update on that as well.\n\nHappy Coding.",
      "votes": 9
    },
    {
      "id": 1316574,
      "postDate": "2021-05-20T17:10:28.433Z",
      "content": "<p>Thanks for sharing! How do you know that this notebook would result in 0.98 on the LB after ensembling as your score on LB is lower than 0.98 now?</p>",
      "rawMarkdown": "Thanks for sharing! How do you know that this notebook would result in 0.98 on the LB after ensembling as your score on LB is lower than 0.98 now?",
      "replies": [
        {
          "id": 1316584,
          "postDate": "2021-05-20T17:27:07.270Z",
          "content": "<p>I said, you can experiment on this to achieve 0.98. I didn't try 5 Fold because of less time. But i guess it should achieve 0.98 based on discussions on CV vs LB topic. </p>",
          "rawMarkdown": "I said, you can experiment on this to achieve 0.98. I didn't try 5 Fold because of less time. But i guess it should achieve 0.98 based on discussions on CV vs LB topic. "
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 1316574,
      "author_name": "Oleg Panichev",
      "author_url": "",
      "post_date": "2021-05-20T17:10:28.433000",
      "content": "<p>Thanks for sharing! How do you know that this notebook would result in 0.98 on the LB after ensembling as your score on LB is lower than 0.98 now?</p>",
      "votes": 0,
      "replies": [
        {
          "id": 1316584,
          "author_name": "Salman",
          "author_url": "",
          "post_date": "2021-05-20T17:27:07.270000",
          "content": "<p>I said, you can experiment on this to achieve 0.98. I didn't try 5 Fold because of less time. But i guess it should achieve 0.98 based on discussions on CV vs LB topic. </p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1316565": "- I have shared EfficientNet B4 trained with MixUP for 10 Epochs achieving 0.97 LB.\n\n- I shared training function, way to use MixUp in [Notebook](https://www.kaggle.com/micheomaano/efficientnet-b4-mixup-cv-0-98-lb-0-97)\n\n- This notebook can be used for experiments on local machines to train a 5 fold to achieve 0.98.\n\nI am working on signal processing to mix some features with images. I will share update on that as well.\n\nHappy Coding.",
    "1316574": "Thanks for sharing! How do you know that this notebook would result in 0.98 on the LB after ensembling as your score on LB is lower than 0.98 now?"
  }
}