{
  "id": 507068,
  "title": "Is training with BCEWithLogitsLoss a good idea? ",
  "url": "/competitions/birdclef-2024/discussion/507068",
  "author_name": "Arindam Roy",
  "post_date": "2024-05-24T09:48:31.828000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hey!</p>\n<p>I am facing a peculiar problem. When my train loss is dropping, my validation loss starts shooting up. However my average validation precision also starts rising. I am not quite sure how both of them  can happen concurrently. I wanted to ask, is my loss function inappropriate? </p>",
  "messages": [
    {
      "id": 2833566,
      "postDate": "2024-05-24T09:48:31.827Z",
      "content": "<p>Hey!</p>\n<p>I am facing a peculiar problem. When my train loss is dropping, my validation loss starts shooting up. However my average validation precision also starts rising. I am not quite sure how both of them  can happen concurrently. I wanted to ask, is my loss function inappropriate? </p>",
      "rawMarkdown": "Hey!\n\nI am facing a peculiar problem. When my train loss is dropping, my validation loss starts shooting up. However my average validation precision also starts rising. I am not quite sure how both of them  can happen concurrently. I wanted to ask, is my loss function inappropriate? \n"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2833566": "Hey!\n\nI am facing a peculiar problem. When my train loss is dropping, my validation loss starts shooting up. However my average validation precision also starts rising. I am not quite sure how both of them  can happen concurrently. I wanted to ask, is my loss function inappropriate? \n"
  }
}