{
  "id": 448925,
  "title": "79th place Solution",
  "content": "<p>Thank you for hosting this competition. <br>\nI'm so excited to have finished my first Kaggle competition with a silver medal!<br>\nThanks to shake up we made our own medal.</p>\n<p>The main techiques are not different as other well organized public notebooks. <br>\nWe took our baseline from the public notebook made by <a href=\"https://www.kaggle.com/tsunotsuno\" target=\"_blank\">@tsunotsuno</a> .<br>\nSpecial thanks to the best contributor of this competition.</p>\n<p>We made a little changes though, which are addressed below briefly.</p>\n<h1>What Worked</h1>\n<ul>\n<li>deberta-v3-large + LGBM</li>\n<li>re-init last 2 layers</li>\n<li>replace '\\n\\n', '\\r\\n' with [PARAGRAPH]</li>\n<li>Use Optuna</li>\n<li>nelder-mead for finding Ensemble weights</li>\n<li>Get target-wised Model</li>\n<li>Post-processing(low impact, private score : + 0.0001)<br>\nWe took a look at the preprocessed data and found some patterns from it.<br>\nSo we arranged the content and wording score like this based on what we analyzed</li>\n</ul>\n<pre><code>test.loc[(test[]&gt;)&amp;(test[]&gt;-), ] = -\ntest.loc[(test[]&gt;)&amp;(test[]&lt;=)&amp;(test[]&gt;), ] = \ntest.loc[(test[]&gt;)&amp;(test[]&gt;-), ] = -\ntest.loc[(test[]&gt;)&amp;(test[]&lt;=)&amp;(test[]&gt;-), ] = -\ntest.loc[(test[]&gt;)&amp;(test[]&lt;=)&amp;(test[]&gt;), ] = \n</code></pre>\n<h1>What Didn't Work</h1>\n<ul>\n<li>freezing n layers</li>\n<li>MeanPooling, AttentionPooling</li>\n<li>change batch size</li>\n<li>Models other than deberta-v3-large (deberta-v2-xlarge, albert-large-v2, muppet-roberta-large, bigbird-roberta-large, roberta-large)</li>\n</ul>\n<p>Thanks again to the competition hosts and everyone who participated!! Good luck for your next competitions!</p>\n<h2><strong>Team member</strong></h2>\n<p><a href=\"https://www.kaggle.com/kimseunghee\" target=\"_blank\">@kimseunghee</a> <br>\n<a href=\"https://www.kaggle.com/danielchae\" target=\"_blank\">@danielchae</a> </p>",
  "comments": [],
  "comment_count": 0,
  "messages": [
    {
      "id": 2491989,
      "postDate": "2023-10-22T07:29:53.917Z",
      "content": "<p>Thank you for hosting this competition. <br>\nI'm so excited to have finished my first Kaggle competition with a silver medal!<br>\nThanks to shake up we made our own medal.</p>\n<p>The main techiques are not different as other well organized public notebooks. <br>\nWe took our baseline from the public notebook made by <a href=\"https://www.kaggle.com/tsunotsuno\" target=\"_blank\">@tsunotsuno</a> .<br>\nSpecial thanks to the best contributor of this competition.</p>\n<p>We made a little changes though, which are addressed below briefly.</p>\n<h1>What Worked</h1>\n<ul>\n<li>deberta-v3-large + LGBM</li>\n<li>re-init last 2 layers</li>\n<li>replace '\\n\\n', '\\r\\n' with [PARAGRAPH]</li>\n<li>Use Optuna</li>\n<li>nelder-mead for finding Ensemble weights</li>\n<li>Get target-wised Model</li>\n<li>Post-processing(low impact, private score : + 0.0001)<br>\nWe took a look at the preprocessed data and found some patterns from it.<br>\nSo we arranged the content and wording score like this based on what we analyzed</li>\n</ul>\n<pre><code>test.loc[(test[]&gt;)&amp;(test[]&gt;-), ] = -\ntest.loc[(test[]&gt;)&amp;(test[]&lt;=)&amp;(test[]&gt;), ] = \ntest.loc[(test[]&gt;)&amp;(test[]&gt;-), ] = -\ntest.loc[(test[]&gt;)&amp;(test[]&lt;=)&amp;(test[]&gt;-), ] = -\ntest.loc[(test[]&gt;)&amp;(test[]&lt;=)&amp;(test[]&gt;), ] = \n</code></pre>\n<h1>What Didn't Work</h1>\n<ul>\n<li>freezing n layers</li>\n<li>MeanPooling, AttentionPooling</li>\n<li>change batch size</li>\n<li>Models other than deberta-v3-large (deberta-v2-xlarge, albert-large-v2, muppet-roberta-large, bigbird-roberta-large, roberta-large)</li>\n</ul>\n<p>Thanks again to the competition hosts and everyone who participated!! Good luck for your next competitions!</p>\n<h2><strong>Team member</strong></h2>\n<p><a href=\"https://www.kaggle.com/kimseunghee\" target=\"_blank\">@kimseunghee</a> <br>\n<a href=\"https://www.kaggle.com/danielchae\" target=\"_blank\">@danielchae</a> </p>",
      "rawMarkdown": "Thank you for hosting this competition. \nI'm so excited to have finished my first Kaggle competition with a silver medal!\nThanks to shake up we made our own medal.\n\nThe main techiques are not different as other well organized public notebooks. \nWe took our baseline from the public notebook made by @tsunotsuno .\nSpecial thanks to the best contributor of this competition.\n\nWe made a little changes though, which are addressed below briefly.\n\n\n# What Worked\n- deberta-v3-large + LGBM\n- re-init last 2 layers\n- replace '\\n\\n', '\\r\\n' with [PARAGRAPH]\n- Use Optuna\n- nelder-mead for finding Ensemble weights\n- Get target-wised Model\n- Post-processing(low impact, private score : + 0.0001)\nWe took a look at the preprocessed data and found some patterns from it.\nSo we arranged the content and wording score like this based on what we analyzed\n\n\n```python\ntest.loc[(test[\"bigram_overlap_ratio\"]>0.9)&(test[\"content\"]>-0.3), \"content\"] = -0.3\ntest.loc[(test[\"bigram_overlap_ratio\"]>0.875)&(test[\"bigram_overlap_ratio\"]<=0.9)&(test[\"content\"]>0), \"content\"] = 0\ntest.loc[(test[\"bigram_overlap_ratio\"]>0.95)&(test[\"wording\"]>-1.1), \"wording\"] = -1.1\ntest.loc[(test[\"bigram_overlap_ratio\"]>0.873)&(test[\"bigram_overlap_ratio\"]<=0.95)&(test[\"wording\"]>-1), \"wording\"] = -1\ntest.loc[(test[\"bigram_overlap_ratio\"]>0.75)&(test[\"bigram_overlap_ratio\"]<=0.873)&(test[\"wording\"]>0), \"wording\"] = 0\n```\n\n# What Didn't Work\n- freezing n layers\n- MeanPooling, AttentionPooling\n- change batch size\n- Models other than deberta-v3-large (deberta-v2-xlarge, albert-large-v2, muppet-roberta-large, bigbird-roberta-large, roberta-large)\n\n\n\nThanks again to the competition hosts and everyone who participated!! Good luck for your next competitions!\n\n\n## **Team member**\n@kimseunghee \n@danielchae ",
      "votes": 6
    }
  ],
  "complete": true,
  "source": "live",
  "fetched_at": "2026-10-19T23:08:39Z",
  "competition_id": 53482,
  "competition_slug": "commonlit-evaluate-student-summaries",
  "team_id": 10725558,
  "solution_url": "/competitions/commonlit-evaluate-student-summaries/writeups/jeonjanggodgame-79th-place-solution"
}