{
  "id": 323838,
  "title": "Data Augmentation is not helping in the LB score.",
  "url": "/competitions/sorghum-id-fgvc-9/discussion/323838",
  "author_name": "",
  "post_date": "2022-05-08T17:24:24.889940800Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Default data augmentation settings for fast ai is not helping in the LB score. I tried ResNet34 model for 5 epochs without data augmentation and then I tried with data augmentation but LB score was low in case of Data Augmentation. Any reason for this? Or am I doing any mistake.</p>\n<p><a href=\"https://www.kaggle.com/code/pksx01/sorghum-level-up-using-fastai\" target=\"_blank\">CV without data augmentation using fastai</a><br>\n<a href=\"https://www.kaggle.com/code/pksx01/sorghum-cultivar-prediction-baseline\" target=\"_blank\">Baseline model using fastai</a></p>",
  "messages": [
    {
      "id": "1781590",
      "postDate": "05/08/2022 17:24:24",
      "content": "<p>Default data augmentation settings for fast ai is not helping in the LB score. I tried ResNet34 model for 5 epochs without data augmentation and then I tried with data augmentation but LB score was low in case of Data Augmentation. Any reason for this? Or am I doing any mistake.</p>\n<p><a href=\"https://www.kaggle.com/code/pksx01/sorghum-level-up-using-fastai\" target=\"_blank\">CV without data augmentation using fastai</a><br>\n<a href=\"https://www.kaggle.com/code/pksx01/sorghum-cultivar-prediction-baseline\" target=\"_blank\">Baseline model using fastai</a></p>",
      "rawMarkdown": "Default data augmentation settings for fast ai is not helping in the LB score. I tried ResNet34 model for 5 epochs without data augmentation and then I tried with data augmentation but LB score was low in case of Data Augmentation. Any reason for this? Or am I doing any mistake.\n\n[CV without data augmentation using fastai](https://www.kaggle.com/code/pksx01/sorghum-level-up-using-fastai)\n[Baseline model using fastai](https://www.kaggle.com/code/pksx01/sorghum-cultivar-prediction-baseline)",
      "votes": null
    },
    {
      "id": "1782641",
      "postDate": "05/09/2022 18:28:37",
      "content": "<p>This is a really interesting question - it's possible that the data augmentation is actually harming your score, rather than helping it. I would recommend trying a few different settings and seeing if you can find a sweet spot. Alternatively, you could try training on less data with data augmentation to see if that makes a difference.</p>",
      "rawMarkdown": "This is a really interesting question - it's possible that the data augmentation is actually harming your score, rather than helping it. I would recommend trying a few different settings and seeing if you can find a sweet spot. Alternatively, you could try training on less data with data augmentation to see if that makes a difference.",
      "votes": null
    },
    {
      "id": "1784258",
      "postDate": "05/11/2022 04:15:38",
      "content": "<p>Thanks! I'll try as per your suggestions. BTW are you looking for a team mate for this competition?</p>",
      "rawMarkdown": "Thanks! I'll try as per your suggestions. BTW are you looking for a team mate for this competition?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1782641,
      "author_name": "",
      "author_url": "",
      "post_date": "05/09/2022 18:28:37",
      "content": "<p>This is a really interesting question - it's possible that the data augmentation is actually harming your score, rather than helping it. I would recommend trying a few different settings and seeing if you can find a sweet spot. Alternatively, you could try training on less data with data augmentation to see if that makes a difference.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1784258,
          "author_name": "pksx01",
          "author_url": "",
          "post_date": "05/11/2022 04:15:38",
          "content": "<p>Thanks! I'll try as per your suggestions. BTW are you looking for a team mate for this competition?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1781590": "Default data augmentation settings for fast ai is not helping in the LB score. I tried ResNet34 model for 5 epochs without data augmentation and then I tried with data augmentation but LB score was low in case of Data Augmentation. Any reason for this? Or am I doing any mistake.\n\n[CV without data augmentation using fastai](https://www.kaggle.com/code/pksx01/sorghum-level-up-using-fastai)\n[Baseline model using fastai](https://www.kaggle.com/code/pksx01/sorghum-cultivar-prediction-baseline)",
    "1782641": "This is a really interesting question - it's possible that the data augmentation is actually harming your score, rather than helping it. I would recommend trying a few different settings and seeing if you can find a sweet spot. Alternatively, you could try training on less data with data augmentation to see if that makes a difference.",
    "1784258": "Thanks! I'll try as per your suggestions. BTW are you looking for a team mate for this competition?"
  },
  "source": "meta"
}