{
  "id": 549015,
  "title": "Both Validation scores are 0.00000 & Notebook Threw Exception",
  "url": "/competitions/ariel-data-challenge-2024/discussion/549015",
  "author_name": "mummyee",
  "post_date": "2024-11-30T08:36:24.236000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi,<br>\nI'm totally new to this notebook submitting competition, and I'm struggling with this problem FOR WEEKS.</p>\n<p>I first preprocessed train/testset, and created two models based on the <a href=\"https://www.kaggle.com/code/gordonyip/host-starter-solution/notebook#Combine-1D-and-2D-CNN-output-for-FINAL-SPECTRA\" target=\"_blank\">HOST START SOLUTION</a> notebook, loaded them on my private dataset.</p>\n<p>Then I created <a href=\"https://www.kaggle.com/code/mummyee/arielsubmission\" target=\"_blank\">my submission file</a> - which predicts, and calculates the uncertainty and creates <code>submission.csv</code> as a submission file.<br>\nAt first I filled my sigma values in 0 to see the logloss of my wl prediction, but got 0.0000 on my private/public score every time I submit my notebook.<br>\nSo I adapted my uncertainty calculating code - and now got 'Notebook Threw Exception' error.</p>\n<p>I searched for some of the others' submissions and it seems like my prediction result is not that different from theirs', so I can't find out why.</p>\n<p>Could anyone please help me find the problem? Or is it just that my model's performance is hideous?</p>",
  "messages": [
    {
      "id": 3059032,
      "postDate": "2024-11-30T08:36:24.237Z",
      "content": "<p>Hi,<br>\nI'm totally new to this notebook submitting competition, and I'm struggling with this problem FOR WEEKS.</p>\n<p>I first preprocessed train/testset, and created two models based on the <a href=\"https://www.kaggle.com/code/gordonyip/host-starter-solution/notebook#Combine-1D-and-2D-CNN-output-for-FINAL-SPECTRA\" target=\"_blank\">HOST START SOLUTION</a> notebook, loaded them on my private dataset.</p>\n<p>Then I created <a href=\"https://www.kaggle.com/code/mummyee/arielsubmission\" target=\"_blank\">my submission file</a> - which predicts, and calculates the uncertainty and creates <code>submission.csv</code> as a submission file.<br>\nAt first I filled my sigma values in 0 to see the logloss of my wl prediction, but got 0.0000 on my private/public score every time I submit my notebook.<br>\nSo I adapted my uncertainty calculating code - and now got 'Notebook Threw Exception' error.</p>\n<p>I searched for some of the others' submissions and it seems like my prediction result is not that different from theirs', so I can't find out why.</p>\n<p>Could anyone please help me find the problem? Or is it just that my model's performance is hideous?</p>",
      "rawMarkdown": "Hi,\nI'm totally new to this notebook submitting competition, and I'm struggling with this problem FOR WEEKS.\n\nI first preprocessed train/testset, and created two models based on the [HOST START SOLUTION](https://www.kaggle.com/code/gordonyip/host-starter-solution/notebook#Combine-1D-and-2D-CNN-output-for-FINAL-SPECTRA) notebook, loaded them on my private dataset.\n\nThen I created [my submission file](https://www.kaggle.com/code/mummyee/arielsubmission) - which predicts, and calculates the uncertainty and creates `submission.csv` as a submission file.\nAt first I filled my sigma values in 0 to see the logloss of my wl prediction, but got 0.0000 on my private/public score every time I submit my notebook.\nSo I adapted my uncertainty calculating code - and now got 'Notebook Threw Exception' error.\n\nI searched for some of the others' submissions and it seems like my prediction result is not that different from theirs', so I can't find out why.\n\nCould anyone please help me find the problem? Or is it just that my model's performance is hideous?"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3059032": "Hi,\nI'm totally new to this notebook submitting competition, and I'm struggling with this problem FOR WEEKS.\n\nI first preprocessed train/testset, and created two models based on the [HOST START SOLUTION](https://www.kaggle.com/code/gordonyip/host-starter-solution/notebook#Combine-1D-and-2D-CNN-output-for-FINAL-SPECTRA) notebook, loaded them on my private dataset.\n\nThen I created [my submission file](https://www.kaggle.com/code/mummyee/arielsubmission) - which predicts, and calculates the uncertainty and creates `submission.csv` as a submission file.\nAt first I filled my sigma values in 0 to see the logloss of my wl prediction, but got 0.0000 on my private/public score every time I submit my notebook.\nSo I adapted my uncertainty calculating code - and now got 'Notebook Threw Exception' error.\n\nI searched for some of the others' submissions and it seems like my prediction result is not that different from theirs', so I can't find out why.\n\nCould anyone please help me find the problem? Or is it just that my model's performance is hideous?"
  }
}