{
  "id": 248252,
  "title": "73rd place solution",
  "url": "/competitions/coleridgeinitiative-show-us-the-data/writeups/let-s-find-the-dataset-73rd-place-solution",
  "author_name": "",
  "post_date": "2021-06-23T00:55:13.523Z",
  "votes": 9,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Our team used an Ensemble of Roberta base and Bert base uncased</p>\n<ul>\n<li>Train data - the competition training data</li>\n<li>Label - Competition labels + extra labels shared by <a href=\"https://www.kaggle.com/mlconsult\" target=\"_blank\">@mlconsult</a> + labels from Paperswithcode.</li>\n</ul>\n<p>Approach</p>\n<ul>\n<li>Roberta model trained for 2 epoch and Bert base trained for 1 Epoch. </li>\n<li>Get common predictions from both the models (intersect the predictions).</li>\n<li>Remove predictions which are &lt;= 2 words other than these ('adni', 'cccsl', 'ibtracs', 'slosh model')</li>\n<li>Along with the predictions add the String match of train_labels and extra labels shared by <a href=\"https://www.kaggle.com/mlconsult\" target=\"_blank\">@mlconsult</a>. </li>\n</ul>\n<p>Apparently the last submission I made for this competition turned out to be the best one that's non-string matching in our private LB too. <br>\nWe were sure that string match won't help in Private LB. But at the same time we were sure if we are performing bad on the public LB with string match it will affect the private LB too, since it is an indication of the possible datasets present in the private LB and did not feel that it could be overfitting the public LB. </p>\n<p>Our intuitions with string match:</p>\n<ul>\n<li><p>Of course like everyone else even my best on the public LB was a string match however I used the string match submission to understand the labels.</p></li>\n<li><p>if we add one extra label and if the score goes up which means the label could be present in the Private LB too and can use that label for training. </p></li>\n<li><p>However if the extra labels though it might sound like a dataset and though it might be in the train data(by string match) and if it does not decreased the public LB this means that the label would a FP since it may not be labelled by the Kaggle team probably. So there is no point in adding them to the model as they may not be labelled in the private LB too. </p></li>\n<li><p>By adding one extra label which sounds like a dataset, if the string match did not change the public LB score/Improve then we can add them to the model. Reason the extra label did not match so no chance of FP on 18% but could be present in private LB, if the score improved then the label is in 18% of the data and chances of being in private data is also high.</p></li>\n</ul>\n<p>Apparently I trusted the Public LB even if it's only 18% of the overall data.</p>\n<p>Thank you so much to my teamate <a href=\"https://www.kaggle.com/jagadish13\" target=\"_blank\">@jagadish13</a> for pairing up with me and chose the final submission trusting the extra label :)</p>\n<p>This competition will be one of the special ones for me though there were lot of changes that happened in between, with not enough labels, no proper validation strategy. </p>\n<p>I got my first medal and I'm happy though I slipped by some 30 positions :)</p>",
  "messages": [
    {
      "id": "1361620",
      "postDate": "06/23/2021 00:50:08",
      "content": "<p>Our team used an Ensemble of Roberta base and Bert base uncased</p>\n<ul>\n<li>Train data - the competition training data</li>\n<li>Label - Competition labels + extra labels shared by <a href=\"https://www.kaggle.com/mlconsult\" target=\"_blank\">@mlconsult</a> + labels from Paperswithcode.</li>\n</ul>\n<p>Approach</p>\n<ul>\n<li>Roberta model trained for 2 epoch and Bert base trained for 1 Epoch. </li>\n<li>Get common predictions from both the models (intersect the predictions).</li>\n<li>Remove predictions which are &lt;= 2 words other than these ('adni', 'cccsl', 'ibtracs', 'slosh model')</li>\n<li>Along with the predictions add the String match of train_labels and extra labels shared by <a href=\"https://www.kaggle.com/mlconsult\" target=\"_blank\">@mlconsult</a>. </li>\n</ul>\n<p>Apparently the last submission I made for this competition turned out to be the best one that's non-string matching in our private LB too. <br>\nWe were sure that string match won't help in Private LB. But at the same time we were sure if we are performing bad on the public LB with string match it will affect the private LB too, since it is an indication of the possible datasets present in the private LB and did not feel that it could be overfitting the public LB. </p>\n<p>Our intuitions with string match:</p>\n<ul>\n<li><p>Of course like everyone else even my best on the public LB was a string match however I used the string match submission to understand the labels.</p></li>\n<li><p>if we add one extra label and if the score goes up which means the label could be present in the Private LB too and can use that label for training. </p></li>\n<li><p>However if the extra labels though it might sound like a dataset and though it might be in the train data(by string match) and if it does not decreased the public LB this means that the label would a FP since it may not be labelled by the Kaggle team probably. So there is no point in adding them to the model as they may not be labelled in the private LB too. </p></li>\n<li><p>By adding one extra label which sounds like a dataset, if the string match did not change the public LB score/Improve then we can add them to the model. Reason the extra label did not match so no chance of FP on 18% but could be present in private LB, if the score improved then the label is in 18% of the data and chances of being in private data is also high.</p></li>\n</ul>\n<p>Apparently I trusted the Public LB even if it's only 18% of the overall data.</p>\n<p>Thank you so much to my teamate <a href=\"https://www.kaggle.com/jagadish13\" target=\"_blank\">@jagadish13</a> for pairing up with me and chose the final submission trusting the extra label :)</p>\n<p>This competition will be one of the special ones for me though there were lot of changes that happened in between, with not enough labels, no proper validation strategy. </p>\n<p>I got my first medal and I'm happy though I slipped by some 30 positions :)</p>",
      "rawMarkdown": "Our team used an Ensemble of Roberta base and Bert base uncased\n- Train data - the competition training data\n- Label - Competition labels + extra labels shared by @mlconsult + labels from Paperswithcode.\n\nApproach\n- Roberta model trained for 2 epoch and Bert base trained for 1 Epoch. \n- Get common predictions from both the models (intersect the predictions).\n- Remove predictions which are <= 2 words other than these ('adni', 'cccsl', 'ibtracs', 'slosh model')\n- Along with the predictions add the String match of train_labels and extra labels shared by @mlconsult. \n\nApparently the last submission I made for this competition turned out to be the best one that's non-string matching in our private LB too. \nWe were sure that string match won't help in Private LB. But at the same time we were sure if we are performing bad on the public LB with string match it will affect the private LB too, since it is an indication of the possible datasets present in the private LB and did not feel that it could be overfitting the public LB. \n\nOur intuitions with string match:\n- Of course like everyone else even my best on the public LB was a string match however I used the string match submission to understand the labels.\n\n- if we add one extra label and if the score goes up which means the label could be present in the Private LB too and can use that label for training. \n\n- However if the extra labels though it might sound like a dataset and though it might be in the train data(by string match) and if it does not decreased the public LB this means that the label would a FP since it may not be labelled by the Kaggle team probably. So there is no point in adding them to the model as they may not be labelled in the private LB too. \n\n- By adding one extra label which sounds like a dataset, if the string match did not change the public LB score/Improve then we can add them to the model. Reason the extra label did not match so no chance of FP on 18% but could be present in private LB, if the score improved then the label is in 18% of the data and chances of being in private data is also high.\n\nApparently I trusted the Public LB even if it's only 18% of the overall data.\n\nThank you so much to my teamate @jagadish13 for pairing up with me and chose the final submission trusting the extra label :)\n\nThis competition will be one of the special ones for me though there were lot of changes that happened in between, with not enough labels, no proper validation strategy. \n\nI got my first medal and I'm happy though I slipped by some 30 positions :)",
      "votes": null
    },
    {
      "id": "1361690",
      "postDate": "06/23/2021 02:41:56",
      "content": "<p>Thanks for sharing your insights</p>",
      "rawMarkdown": "Thanks for sharing your insights",
      "votes": null
    },
    {
      "id": "1361700",
      "postDate": "06/23/2021 02:53:15",
      "content": "<p>You are welcome 👍 😊</p>",
      "rawMarkdown": "You are welcome 👍 😊",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1361690,
      "author_name": "ssinyu",
      "author_url": "",
      "post_date": "06/23/2021 02:41:56",
      "content": "<p>Thanks for sharing your insights</p>",
      "votes": null,
      "replies": [
        {
          "id": 1361700,
          "author_name": "thanish",
          "author_url": "",
          "post_date": "06/23/2021 02:53:15",
          "content": "<p>You are welcome 👍 😊</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1361620": "Our team used an Ensemble of Roberta base and Bert base uncased\n- Train data - the competition training data\n- Label - Competition labels + extra labels shared by @mlconsult + labels from Paperswithcode.\n\nApproach\n- Roberta model trained for 2 epoch and Bert base trained for 1 Epoch. \n- Get common predictions from both the models (intersect the predictions).\n- Remove predictions which are <= 2 words other than these ('adni', 'cccsl', 'ibtracs', 'slosh model')\n- Along with the predictions add the String match of train_labels and extra labels shared by @mlconsult. \n\nApparently the last submission I made for this competition turned out to be the best one that's non-string matching in our private LB too. \nWe were sure that string match won't help in Private LB. But at the same time we were sure if we are performing bad on the public LB with string match it will affect the private LB too, since it is an indication of the possible datasets present in the private LB and did not feel that it could be overfitting the public LB. \n\nOur intuitions with string match:\n- Of course like everyone else even my best on the public LB was a string match however I used the string match submission to understand the labels.\n\n- if we add one extra label and if the score goes up which means the label could be present in the Private LB too and can use that label for training. \n\n- However if the extra labels though it might sound like a dataset and though it might be in the train data(by string match) and if it does not decreased the public LB this means that the label would a FP since it may not be labelled by the Kaggle team probably. So there is no point in adding them to the model as they may not be labelled in the private LB too. \n\n- By adding one extra label which sounds like a dataset, if the string match did not change the public LB score/Improve then we can add them to the model. Reason the extra label did not match so no chance of FP on 18% but could be present in private LB, if the score improved then the label is in 18% of the data and chances of being in private data is also high.\n\nApparently I trusted the Public LB even if it's only 18% of the overall data.\n\nThank you so much to my teamate @jagadish13 for pairing up with me and chose the final submission trusting the extra label :)\n\nThis competition will be one of the special ones for me though there were lot of changes that happened in between, with not enough labels, no proper validation strategy. \n\nI got my first medal and I'm happy though I slipped by some 30 positions :)",
    "1361690": "Thanks for sharing your insights",
    "1361700": "You are welcome 👍 😊"
  },
  "source": "meta"
}