{
  "id": 236647,
  "title": "eval score GAP of single model",
  "url": "/competitions/coleridgeinitiative-show-us-the-data/discussion/236647",
  "author_name": "van73",
  "post_date": "2021-05-05T08:49:34.785000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Dear friends, what are the online and offline scores of your single model? I am puzzled by the huge difference between the scores of my own verification set and the scores of the public test set.</p>\n<p>The multi-response QA model I used uses the evaluation method from here: <br>\n<a href=\"url\" target=\"_blank\">https://www.kaggle.com/c/coleridgeinitiative-show-us-the-data/discussion/230091</a></p>\n<p>And 80% of the data is used as the training set, and 20% of the data is used as the validation set. The score of the validation set is <strong>0.848</strong>, but only <strong>0.32</strong> online!</p>\n<p>So does the fusion of multi-fold models have a great impact on the scores? Dear friends, can speak freely</p>",
  "messages": [
    {
      "id": 1293904,
      "postDate": "2021-05-05T08:49:34.787Z",
      "content": "<p>Dear friends, what are the online and offline scores of your single model? I am puzzled by the huge difference between the scores of my own verification set and the scores of the public test set.</p>\n<p>The multi-response QA model I used uses the evaluation method from here: <br>\n<a href=\"url\" target=\"_blank\">https://www.kaggle.com/c/coleridgeinitiative-show-us-the-data/discussion/230091</a></p>\n<p>And 80% of the data is used as the training set, and 20% of the data is used as the validation set. The score of the validation set is <strong>0.848</strong>, but only <strong>0.32</strong> online!</p>\n<p>So does the fusion of multi-fold models have a great impact on the scores? Dear friends, can speak freely</p>",
      "rawMarkdown": "Dear friends, what are the online and offline scores of your single model? I am puzzled by the huge difference between the scores of my own verification set and the scores of the public test set.\n\nThe multi-response QA model I used uses the evaluation method from here: \n[https://www.kaggle.com/c/coleridgeinitiative-show-us-the-data/discussion/230091](url)\n\nAnd 80% of the data is used as the training set, and 20% of the data is used as the validation set. The score of the validation set is **0.848**, but only **0.32** online!\n\nSo does the fusion of multi-fold models have a great impact on the scores? Dear friends, can speak freely\n\n\n\n",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1293904": "Dear friends, what are the online and offline scores of your single model? I am puzzled by the huge difference between the scores of my own verification set and the scores of the public test set.\n\nThe multi-response QA model I used uses the evaluation method from here: \n[https://www.kaggle.com/c/coleridgeinitiative-show-us-the-data/discussion/230091](url)\n\nAnd 80% of the data is used as the training set, and 20% of the data is used as the validation set. The score of the validation set is **0.848**, but only **0.32** online!\n\nSo does the fusion of multi-fold models have a great impact on the scores? Dear friends, can speak freely\n\n\n\n"
  }
}