{
  "id": 168727,
  "title": "Score not improving.",
  "url": "/competitions/global-wheat-detection/discussion/168727",
  "author_name": "",
  "post_date": "2020-07-21T16:20:33.046972700Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Hey, I am using a retinanet model for this competition and I see that how much ever I train my model my score is not improving. I tried different sizes of image in train data and increasing the epochs. <br>\nBut still, I get the same score in the competition. <br>\nCan anyone please help me understand what is my mistake.<br>\nThank you.</p>",
  "messages": [
    {
      "id": "938608",
      "postDate": "07/21/2020 16:20:33",
      "content": "<p>Hey, I am using a retinanet model for this competition and I see that how much ever I train my model my score is not improving. I tried different sizes of image in train data and increasing the epochs. <br>\nBut still, I get the same score in the competition. <br>\nCan anyone please help me understand what is my mistake.<br>\nThank you.</p>",
      "rawMarkdown": "Hey, I am using a retinanet model for this competition and I see that how much ever I train my model my score is not improving. I tried different sizes of image in train data and increasing the epochs. \nBut still, I get the same score in the competition. \nCan anyone please help me understand what is my mistake.\nThank you.",
      "votes": null
    },
    {
      "id": "938615",
      "postDate": "07/21/2020 16:25:27",
      "content": "<p>Did you tried reducing the learning rate? </p>\n\n<p>How many epochs are you using?</p>",
      "rawMarkdown": "Did you tried reducing the learning rate? \n\nHow many epochs are you using?",
      "votes": null
    },
    {
      "id": "939302",
      "postDate": "07/22/2020 06:45:47",
      "content": "<p>use some augmentations, if you don`t</p>",
      "rawMarkdown": "use some augmentations, if you don`t",
      "votes": null
    },
    {
      "id": "942380",
      "postDate": "07/23/2020 18:04:39",
      "content": "<p>Okay.\nthank you.</p>",
      "rawMarkdown": "Okay.\nthank you.",
      "votes": null
    },
    {
      "id": "942386",
      "postDate": "07/23/2020 18:06:18",
      "content": "<p>I am using adaptive lr.\nI trained up to 10 epochs. But I tried with different sizes of train images</p>",
      "rawMarkdown": "I am using adaptive lr.\nI trained up to 10 epochs. But I tried with different sizes of train images",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 938615,
      "author_name": "luisvivas",
      "author_url": "",
      "post_date": "07/21/2020 16:25:27",
      "content": "<p>Did you tried reducing the learning rate? </p>\n\n<p>How many epochs are you using?</p>",
      "votes": null,
      "replies": [
        {
          "id": 942386,
          "author_name": "saiseshank",
          "author_url": "",
          "post_date": "07/23/2020 18:06:18",
          "content": "<p>I am using adaptive lr.\nI trained up to 10 epochs. But I tried with different sizes of train images</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 939302,
      "author_name": "kupchanski",
      "author_url": "",
      "post_date": "07/22/2020 06:45:47",
      "content": "<p>use some augmentations, if you don`t</p>",
      "votes": null,
      "replies": [
        {
          "id": 942380,
          "author_name": "saiseshank",
          "author_url": "",
          "post_date": "07/23/2020 18:04:39",
          "content": "<p>Okay.\nthank you.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "938608": "Hey, I am using a retinanet model for this competition and I see that how much ever I train my model my score is not improving. I tried different sizes of image in train data and increasing the epochs. \nBut still, I get the same score in the competition. \nCan anyone please help me understand what is my mistake.\nThank you.",
    "938615": "Did you tried reducing the learning rate? \n\nHow many epochs are you using?",
    "939302": "use some augmentations, if you don`t",
    "942380": "Okay.\nthank you.",
    "942386": "I am using adaptive lr.\nI trained up to 10 epochs. But I tried with different sizes of train images"
  },
  "source": "meta"
}