{
  "id": 71766,
  "title": "Some bug of my models",
  "url": "/competitions/quickdraw-doodle-recognition/discussion/71766",
  "author_name": "TPloveYXT520",
  "post_date": "2018-11-16T09:49:21.295000",
  "votes": -2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>When I use the resnet_50 and image_size =128*128\nI get the error that my loss is quickly decrease to the 0 and my top 3 accuracy is also decrease\nbut I do not get the error when i use the mobilenet \nI saw a lot of people use the resnet50 and image_size=128*128, Do you get the error\nI use the keras</p>",
  "messages": [
    {
      "id": 422492,
      "postDate": "2018-11-16T09:49:21.297Z",
      "content": "<p>When I use the resnet_50 and image_size =128*128\nI get the error that my loss is quickly decrease to the 0 and my top 3 accuracy is also decrease\nbut I do not get the error when i use the mobilenet \nI saw a lot of people use the resnet50 and image_size=128*128, Do you get the error\nI use the keras</p>",
      "rawMarkdown": "When I use the resnet_50 and image_size =128*128\nI get the error that my loss is quickly decrease to the 0 and my top 3 accuracy is also decrease\nbut I do not get the error when i use the mobilenet \nI saw a lot of people use the resnet50 and image_size=128*128, Do you get the error\nI use the keras",
      "votes": -2
    },
    {
      "id": 422509,
      "postDate": "2018-11-16T10:29:07.387Z",
      "content": "<p>loss=0?</p>",
      "rawMarkdown": "loss=0?"
    }
  ],
  "comments": [
    {
      "id": 422509,
      "author_name": "Gary",
      "author_url": "",
      "post_date": "2018-11-16T10:29:07.387000",
      "content": "<p>loss=0?</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "422492": "When I use the resnet_50 and image_size =128*128\nI get the error that my loss is quickly decrease to the 0 and my top 3 accuracy is also decrease\nbut I do not get the error when i use the mobilenet \nI saw a lot of people use the resnet50 and image_size=128*128, Do you get the error\nI use the keras",
    "422509": "loss=0?"
  }
}