{
  "id": 277652,
  "title": "\"Submission CSV Not Found\" when converting test dataset to nifti",
  "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/277652",
  "author_name": "qdstro",
  "post_date": "2021-10-10T14:34:28.914000",
  "votes": 3,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Dear Kagglers,</p>\n<p>I am experiencing strange behavior when trying to submit my notebook. Even though the execution of the notebook on the public test set does not throw an error, and a submission.csv is generated, the \"Submissions\" page shows a \"Submission CSV Not Found\" error message.</p>\n<p>I created a notebook to recreate the error: <a href=\"https://www.kaggle.com/qdstro/submission-test\" target=\"_blank\">https://www.kaggle.com/qdstro/submission-test</a></p>\n<p>The submission works when omitting the \"Preprocessing\" step for loading and saving the data. I read that the kernel may be out of memory, but I don't see why and how to reduce RAM usage when loading the single series. I already tried reducing the image resolution before saving with no effect. </p>\n<p>I also tested different alternatives like dicom2nifti and various combinations of try-except statements without success.</p>\n<p>The complete code will run inference using an externally trained model. </p>\n<p>Anybody experiencing something similar? How may I resolve this issue? Thanks!</p>",
  "messages": [
    {
      "id": 1540518,
      "postDate": "2021-10-10T14:34:28.913Z",
      "content": "<p>Dear Kagglers,</p>\n<p>I am experiencing strange behavior when trying to submit my notebook. Even though the execution of the notebook on the public test set does not throw an error, and a submission.csv is generated, the \"Submissions\" page shows a \"Submission CSV Not Found\" error message.</p>\n<p>I created a notebook to recreate the error: <a href=\"https://www.kaggle.com/qdstro/submission-test\" target=\"_blank\">https://www.kaggle.com/qdstro/submission-test</a></p>\n<p>The submission works when omitting the \"Preprocessing\" step for loading and saving the data. I read that the kernel may be out of memory, but I don't see why and how to reduce RAM usage when loading the single series. I already tried reducing the image resolution before saving with no effect. </p>\n<p>I also tested different alternatives like dicom2nifti and various combinations of try-except statements without success.</p>\n<p>The complete code will run inference using an externally trained model. </p>\n<p>Anybody experiencing something similar? How may I resolve this issue? Thanks!</p>",
      "rawMarkdown": "Dear Kagglers,\n\nI am experiencing strange behavior when trying to submit my notebook. Even though the execution of the notebook on the public test set does not throw an error, and a submission.csv is generated, the \"Submissions\" page shows a \"Submission CSV Not Found\" error message.\n\n I created a notebook to recreate the error: https://www.kaggle.com/qdstro/submission-test\n\nThe submission works when omitting the \"Preprocessing\" step for loading and saving the data. I read that the kernel may be out of memory, but I don't see why and how to reduce RAM usage when loading the single series. I already tried reducing the image resolution before saving with no effect. \n\nI also tested different alternatives like dicom2nifti and various combinations of try-except statements without success.\n\nThe complete code will run inference using an externally trained model. \n\nAnybody experiencing something similar? How may I resolve this issue? Thanks!",
      "votes": 3
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1540518": "Dear Kagglers,\n\nI am experiencing strange behavior when trying to submit my notebook. Even though the execution of the notebook on the public test set does not throw an error, and a submission.csv is generated, the \"Submissions\" page shows a \"Submission CSV Not Found\" error message.\n\n I created a notebook to recreate the error: https://www.kaggle.com/qdstro/submission-test\n\nThe submission works when omitting the \"Preprocessing\" step for loading and saving the data. I read that the kernel may be out of memory, but I don't see why and how to reduce RAM usage when loading the single series. I already tried reducing the image resolution before saving with no effect. \n\nI also tested different alternatives like dicom2nifti and various combinations of try-except statements without success.\n\nThe complete code will run inference using an externally trained model. \n\nAnybody experiencing something similar? How may I resolve this issue? Thanks!"
  }
}