{
  "id": 512511,
  "title": "Submission Submission CSV Not Found, something doesn't seem right",
  "url": "/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/512511",
  "author_name": "",
  "post_date": "2024-06-15T14:00:52.789544900Z",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I'm encountering an issue where my code, wrapped in try-except blocks, produces correct outputs with predefined values. However, the Docker machine seems to die silently during the image processing part. I'm using the joblib library to process DCM files, which has worked in many competitions before. I also set the number of jobs to the number of CPUs.</p>\n<p>Is anyone else struggling with a similar issue? Any hints to fix this would be appreciated.</p>\n<p>Another thing I noticed is that the run for submission happened on the \"Latest Container Image\", even though I specified \"Pin to original environment\" in the settings. How can I change this?</p>\n<p>It seems to me that the VM is dying without detection. I'm almost certain this is the case, but maybe not.</p>\n<p>I'm not sure what to do other than putting things in try/except loops. I could change the processing to be on the fly or avoid pre-processing the images altogether, but this would take a long time to submit.</p>\n<p>It seems that in my case, the test environment is significantly different from the submission environment.</p>",
  "messages": [
    {
      "id": "2873439",
      "postDate": "06/15/2024 14:00:52",
      "content": "<p>I'm encountering an issue where my code, wrapped in try-except blocks, produces correct outputs with predefined values. However, the Docker machine seems to die silently during the image processing part. I'm using the joblib library to process DCM files, which has worked in many competitions before. I also set the number of jobs to the number of CPUs.</p>\n<p>Is anyone else struggling with a similar issue? Any hints to fix this would be appreciated.</p>\n<p>Another thing I noticed is that the run for submission happened on the \"Latest Container Image\", even though I specified \"Pin to original environment\" in the settings. How can I change this?</p>\n<p>It seems to me that the VM is dying without detection. I'm almost certain this is the case, but maybe not.</p>\n<p>I'm not sure what to do other than putting things in try/except loops. I could change the processing to be on the fly or avoid pre-processing the images altogether, but this would take a long time to submit.</p>\n<p>It seems that in my case, the test environment is significantly different from the submission environment.</p>",
      "rawMarkdown": "I'm encountering an issue where my code, wrapped in try-except blocks, produces correct outputs with predefined values. However, the Docker machine seems to die silently during the image processing part. I'm using the joblib library to process DCM files, which has worked in many competitions before. I also set the number of jobs to the number of CPUs.\n\nIs anyone else struggling with a similar issue? Any hints to fix this would be appreciated.\n\nAnother thing I noticed is that the run for submission happened on the \"Latest Container Image\", even though I specified \"Pin to original environment\" in the settings. How can I change this?\n\nIt seems to me that the VM is dying without detection. I'm almost certain this is the case, but maybe not.\n\nI'm not sure what to do other than putting things in try/except loops. I could change the processing to be on the fly or avoid pre-processing the images altogether, but this would take a long time to submit.\n\nIt seems that in my case, the test environment is significantly different from the submission environment.",
      "votes": null
    },
    {
      "id": "2873765",
      "postDate": "06/15/2024 18:25:42",
      "content": "<p>My suggestion is to sample, say 200 studies, from train folder and test run on the latest container image (as you think it is being used). That may help in debugging the problem.</p>",
      "rawMarkdown": "My suggestion is to sample, say 200 studies, from train folder and test run on the latest container image (as you think it is being used). That may help in debugging the problem.",
      "votes": null
    },
    {
      "id": "2873771",
      "postDate": "06/15/2024 18:30:11",
      "content": "<p>That's a great suggestion!… Thank you <a href=\"https://www.kaggle.com/coderrkj\" target=\"_blank\">@coderrkj</a> </p>",
      "rawMarkdown": "That's a great suggestion!... Thank you @coderrkj",
      "votes": null
    },
    {
      "id": "2874145",
      "postDate": "06/16/2024 04:45:06",
      "content": "<p>Another thing I noticed is that the run for submission happened on the \"Latest Container Image\", even though I specified \"Pin to original environment\" in the settings. How can I change this?</p>\n<p>Have you tried, restarting and notebook then setting the environment to the original environment by click on the dropdown <a href=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F14118568%2F8de990f9a773bc78d305fb2cd3d08caa%2Fenv.png?generation=1718513061132334&amp;alt=media\" target=\"_blank\">button</a> and then again restarting the notebook?</p>",
      "rawMarkdown": "Another thing I noticed is that the run for submission happened on the \"Latest Container Image\", even though I specified \"Pin to original environment\" in the settings. How can I change this?\n\nHave you tried, restarting and notebook then setting the environment to the original environment by click on the dropdown [button](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F14118568%2F8de990f9a773bc78d305fb2cd3d08caa%2Fenv.png?generation=1718513061132334&alt=media) and then again restarting the notebook?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2873765,
      "author_name": "coderrkj",
      "author_url": "",
      "post_date": "06/15/2024 18:25:42",
      "content": "<p>My suggestion is to sample, say 200 studies, from train folder and test run on the latest container image (as you think it is being used). That may help in debugging the problem.</p>",
      "votes": null,
      "replies": [
        {
          "id": 2873771,
          "author_name": "abualabed",
          "author_url": "",
          "post_date": "06/15/2024 18:30:11",
          "content": "<p>That's a great suggestion!… Thank you <a href=\"https://www.kaggle.com/coderrkj\" target=\"_blank\">@coderrkj</a> </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 2874145,
      "author_name": "devsya",
      "author_url": "",
      "post_date": "06/16/2024 04:45:06",
      "content": "<p>Another thing I noticed is that the run for submission happened on the \"Latest Container Image\", even though I specified \"Pin to original environment\" in the settings. How can I change this?</p>\n<p>Have you tried, restarting and notebook then setting the environment to the original environment by click on the dropdown <a href=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F14118568%2F8de990f9a773bc78d305fb2cd3d08caa%2Fenv.png?generation=1718513061132334&amp;alt=media\" target=\"_blank\">button</a> and then again restarting the notebook?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2873439": "I'm encountering an issue where my code, wrapped in try-except blocks, produces correct outputs with predefined values. However, the Docker machine seems to die silently during the image processing part. I'm using the joblib library to process DCM files, which has worked in many competitions before. I also set the number of jobs to the number of CPUs.\n\nIs anyone else struggling with a similar issue? Any hints to fix this would be appreciated.\n\nAnother thing I noticed is that the run for submission happened on the \"Latest Container Image\", even though I specified \"Pin to original environment\" in the settings. How can I change this?\n\nIt seems to me that the VM is dying without detection. I'm almost certain this is the case, but maybe not.\n\nI'm not sure what to do other than putting things in try/except loops. I could change the processing to be on the fly or avoid pre-processing the images altogether, but this would take a long time to submit.\n\nIt seems that in my case, the test environment is significantly different from the submission environment.",
    "2873765": "My suggestion is to sample, say 200 studies, from train folder and test run on the latest container image (as you think it is being used). That may help in debugging the problem.",
    "2873771": "That's a great suggestion!... Thank you @coderrkj",
    "2874145": "Another thing I noticed is that the run for submission happened on the \"Latest Container Image\", even though I specified \"Pin to original environment\" in the settings. How can I change this?\n\nHave you tried, restarting and notebook then setting the environment to the original environment by click on the dropdown [button](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F14118568%2F8de990f9a773bc78d305fb2cd3d08caa%2Fenv.png?generation=1718513061132334&alt=media) and then again restarting the notebook?"
  },
  "source": "meta"
}