{
  "id": 75445,
  "title": "Fluctuations in Validation set",
  "url": "/competitions/dogs-vs-cats-redux-kernels-edition/discussion/75445",
  "author_name": "",
  "post_date": "2018-12-21T18:08:08.707649200Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>While making a model with CNN for Cats and Dogs recognition I have found my validation set fluctuating heavily. Training and validation accuracy both are good but grapg is fluctuating heavily for validation set. Can anyone suugest me a way around where I am going wrong. Any idea. Using LR= 0.0001, Optimiser= RMSProp, Adam, SGD etc. Shuffle=True, </p>",
  "messages": [
    {
      "id": "443480",
      "postDate": "12/21/2018 18:08:08",
      "content": "<p>While making a model with CNN for Cats and Dogs recognition I have found my validation set fluctuating heavily. Training and validation accuracy both are good but grapg is fluctuating heavily for validation set. Can anyone suugest me a way around where I am going wrong. Any idea. Using LR= 0.0001, Optimiser= RMSProp, Adam, SGD etc. Shuffle=True, </p>",
      "rawMarkdown": "While making a model with CNN for Cats and Dogs recognition I have found my validation set fluctuating heavily. Training and validation accuracy both are good but grapg is fluctuating heavily for validation set. Can anyone suugest me a way around where I am going wrong. Any idea. Using LR= 0.0001, Optimiser= RMSProp, Adam, SGD etc. Shuffle=True,",
      "votes": null
    },
    {
      "id": "461681",
      "postDate": "01/26/2019 17:59:10",
      "content": "<p>Hello All I have solved this problem by using reducing learning rate… I got dramaticaly improved result by using this technique.. Thanx..</p>",
      "rawMarkdown": "Hello All I have solved this problem by using reducing learning rate… I got dramaticaly improved result by using this technique.. Thanx..",
      "votes": null
    },
    {
      "id": "528579",
      "postDate": "05/08/2019 05:32:43",
      "content": "<p>cool glad to help</p>",
      "rawMarkdown": "cool glad to help",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 461681,
      "author_name": "hengulkakaty",
      "author_url": "",
      "post_date": "01/26/2019 17:59:10",
      "content": "<p>Hello All I have solved this problem by using reducing learning rate… I got dramaticaly improved result by using this technique.. Thanx..</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 528579,
      "author_name": "zarifaziz08",
      "author_url": "",
      "post_date": "05/08/2019 05:32:43",
      "content": "<p>cool glad to help</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "443480": "While making a model with CNN for Cats and Dogs recognition I have found my validation set fluctuating heavily. Training and validation accuracy both are good but grapg is fluctuating heavily for validation set. Can anyone suugest me a way around where I am going wrong. Any idea. Using LR= 0.0001, Optimiser= RMSProp, Adam, SGD etc. Shuffle=True,",
    "461681": "Hello All I have solved this problem by using reducing learning rate… I got dramaticaly improved result by using this technique.. Thanx..",
    "528579": "cool glad to help"
  },
  "source": "meta"
}