{
  "id": 79982,
  "title": "What are your final submission local CV scores",
  "url": "/competitions/quora-insincere-questions-classification/discussion/79982",
  "author_name": "",
  "post_date": "2019-02-09T07:38:06.547517400Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Mine were :</p>\n\n<ol>\n<li>0.69 Local CV - Single model - 5 fold CV along with checkpoint ensembling</li>\n<li>0.70 Local CV - Logistic regression Stack of 4 models </li>\n</ol>",
  "messages": [
    {
      "id": "468584",
      "postDate": "02/09/2019 07:38:06",
      "content": "<p>Mine were :</p>\n\n<ol>\n<li>0.69 Local CV - Single model - 5 fold CV along with checkpoint ensembling</li>\n<li>0.70 Local CV - Logistic regression Stack of 4 models </li>\n</ol>",
      "rawMarkdown": "Mine were :\n\n1. 0.69 Local CV - Single model - 5 fold CV along with checkpoint ensembling\n2. 0.70 Local CV - Logistic regression Stack of 4 models",
      "votes": null
    },
    {
      "id": "470135",
      "postDate": "02/12/2019 12:54:53",
      "content": "<p>hi, mine was 0.683.\nI'm a newbie and can u tell me some details to achieve 0.69 local cv?\nI used text preprocessing and lstm+attention+capsule and 5 fold. I had tried some ways but no improvement. Thank u!</p>",
      "rawMarkdown": "hi, mine was 0.683.\nI'm a newbie and can u tell me some details to achieve 0.69 local cv?\nI used text preprocessing and lstm+attention+capsule and 5 fold. I had tried some ways but no improvement. Thank u!",
      "votes": null
    },
    {
      "id": "470255",
      "postDate": "02/12/2019 16:48:10",
      "content": "<p>I created predictions of test data after each epoch and did a weighted ensembling of those predictions to arrive at a final submission where i gave lower weights to start epochs. Gave me a boost from 0.68 to 0.69 Local CV. See this thread: <a href=\"https://www.kaggle.com/c/quora-insincere-questions-classification/discussion/79720\">https://www.kaggle.com/c/quora-insincere-questions-classification/discussion/79720</a></p>",
      "rawMarkdown": "I created predictions of test data after each epoch and did a weighted ensembling of those predictions to arrive at a final submission where i gave lower weights to start epochs. Gave me a boost from 0.68 to 0.69 Local CV. See this thread: https://www.kaggle.com/c/quora-insincere-questions-classification/discussion/79720",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 470135,
      "author_name": "karwik",
      "author_url": "",
      "post_date": "02/12/2019 12:54:53",
      "content": "<p>hi, mine was 0.683.\nI'm a newbie and can u tell me some details to achieve 0.69 local cv?\nI used text preprocessing and lstm+attention+capsule and 5 fold. I had tried some ways but no improvement. Thank u!</p>",
      "votes": null,
      "replies": [
        {
          "id": 470255,
          "author_name": "mlwhiz",
          "author_url": "",
          "post_date": "02/12/2019 16:48:10",
          "content": "<p>I created predictions of test data after each epoch and did a weighted ensembling of those predictions to arrive at a final submission where i gave lower weights to start epochs. Gave me a boost from 0.68 to 0.69 Local CV. See this thread: <a href=\"https://www.kaggle.com/c/quora-insincere-questions-classification/discussion/79720\">https://www.kaggle.com/c/quora-insincere-questions-classification/discussion/79720</a></p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "468584": "Mine were :\n\n1. 0.69 Local CV - Single model - 5 fold CV along with checkpoint ensembling\n2. 0.70 Local CV - Logistic regression Stack of 4 models",
    "470135": "hi, mine was 0.683.\nI'm a newbie and can u tell me some details to achieve 0.69 local cv?\nI used text preprocessing and lstm+attention+capsule and 5 fold. I had tried some ways but no improvement. Thank u!",
    "470255": "I created predictions of test data after each epoch and did a weighted ensembling of those predictions to arrive at a final submission where i gave lower weights to start epochs. Gave me a boost from 0.68 to 0.69 Local CV. See this thread: https://www.kaggle.com/c/quora-insincere-questions-classification/discussion/79720"
  },
  "source": "meta"
}