{
  "id": 170326,
  "title": "Non post-processed LB scores with single fold models? Especially FasterRCNN.",
  "url": "/competitions/global-wheat-detection/discussion/170326",
  "author_name": "",
  "post_date": "2020-07-27T08:41:50.918757600Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>What are your best LB scores in this regard? I can't break into 0.7 so trying to understand how everyone is doing.</p>",
  "messages": [
    {
      "id": "947399",
      "postDate": "07/27/2020 08:41:50",
      "content": "<p>What are your best LB scores in this regard? I can't break into 0.7 so trying to understand how everyone is doing.</p>",
      "rawMarkdown": "What are your best LB scores in this regard? I can't break into 0.7 so trying to understand how everyone is doing.",
      "votes": null
    },
    {
      "id": "951293",
      "postDate": "07/30/2020 03:40:15",
      "content": "<p>Try Faster RCNN with ResNet152 base. It will help. This helped me to get above 0.7 on LB. Although it feels like using bigger networks is helping but the training time is increasing a lot. I don't know how much more I will be able to experiment with 6 days left. Good Luck.</p>",
      "rawMarkdown": "Try Faster RCNN with ResNet152 base. It will help. This helped me to get above 0.7 on LB. Although it feels like using bigger networks is helping but the training time is increasing a lot. I don't know how much more I will be able to experiment with 6 days left. Good Luck.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 951293,
      "author_name": "sovitrath",
      "author_url": "",
      "post_date": "07/30/2020 03:40:15",
      "content": "<p>Try Faster RCNN with ResNet152 base. It will help. This helped me to get above 0.7 on LB. Although it feels like using bigger networks is helping but the training time is increasing a lot. I don't know how much more I will be able to experiment with 6 days left. Good Luck.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "947399": "What are your best LB scores in this regard? I can't break into 0.7 so trying to understand how everyone is doing.",
    "951293": "Try Faster RCNN with ResNet152 base. It will help. This helped me to get above 0.7 on LB. Although it feels like using bigger networks is helping but the training time is increasing a lot. I don't know how much more I will be able to experiment with 6 days left. Good Luck."
  },
  "source": "meta"
}