{
  "id": 371105,
  "title": "Why my sub always Threw Exception??",
  "url": "/competitions/rsna-breast-cancer-detection/discussion/371105",
  "author_name": "wqx2022",
  "post_date": "2022-12-08T01:44:44.563000",
  "votes": 0,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I use this <a href=\"https://www.kaggle.com/wqx2022/smbcd-pytorch-inference-sub\" target=\"_blank\">inference template</a> to generate submission.csv, everything seems normal. But my sub always Threw Exception, I kept my Internet OFF, I processed the DICOM to PNG into /kaggle/temp/XXX like someone's do, I did not use GPUs… It almost drives me mad, could anyone help me plz?</p>",
  "messages": [
    {
      "id": 2058575,
      "postDate": "2022-12-08T03:25:56.727Z",
      "content": "<p>hey <a href=\"https://www.kaggle.com/wqx2022\" target=\"_blank\">@wqx2022</a>!</p>\n<p>There is a chance that missing this is the reason:</p>\n<pre><code>try:\n    import pylibjpeg\nexcept:\n    !rm -rf /root/.cache/torch/hub/checkpoints/\n    !mkdir -p /root/.cache/torch/hub/checkpoints/\n    !pip install /kaggle/input/rsna-2022-whl/{pydicom-2.3.0-py3-none-any.whl,pylibjpeg-1.4.0-py3-none-any.whl,python_gdcm-3.0.15-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl}\n    !pip install /kaggle/input/rsna-2022-whl/{torch-1.12.1-cp37-cp37m-manylinux1_x86_64.whl,torchvision-0.13.1-cp37-cp37m-manylinux1_x86_64.whl}\n</code></pre>\n<p>Without this you will be able to process the files in the visible test set, but not the hidden test set that gets attached to your VM later on.</p>\n<p>You also need to attach the <code>rsna-2022-whl</code> dataset to your notebook.</p>\n<p>BTW a good heuristic to try to identify what might be the issue is trying to predict on the train set. It has more files coming from different machines -- if your submission will run on some part of the train set, it will quite likely run on the hidden test set as well!</p>",
      "rawMarkdown": "hey @wqx2022!\n\nThere is a chance that missing this is the reason:\n\n```\ntry:\n    import pylibjpeg\nexcept:\n    !rm -rf /root/.cache/torch/hub/checkpoints/\n    !mkdir -p /root/.cache/torch/hub/checkpoints/\n    !pip install /kaggle/input/rsna-2022-whl/{pydicom-2.3.0-py3-none-any.whl,pylibjpeg-1.4.0-py3-none-any.whl,python_gdcm-3.0.15-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl}\n    !pip install /kaggle/input/rsna-2022-whl/{torch-1.12.1-cp37-cp37m-manylinux1_x86_64.whl,torchvision-0.13.1-cp37-cp37m-manylinux1_x86_64.whl}\n```\nWithout this you will be able to process the files in the visible test set, but not the hidden test set that gets attached to your VM later on.\n\nYou also need to attach the `rsna-2022-whl` dataset to your notebook.\n\nBTW a good heuristic to try to identify what might be the issue is trying to predict on the train set. It has more files coming from different machines -- if your submission will run on some part of the train set, it will quite likely run on the hidden test set as well!",
      "votes": 4,
      "replies": [
        {
          "id": 2058581,
          "postDate": "2022-12-08T03:46:09.817Z",
          "content": "<p>Yeah, thx, I'll try it later.</p>",
          "rawMarkdown": "Yeah, thx, I'll try it later.",
          "replies": [
            {
              "id": 2125651,
              "postDate": "2023-02-01T19:14:35.593Z",
              "content": "<p>how did you solve your problem of notebook throwing exception on submission?</p>",
              "rawMarkdown": "how did you solve your problem of notebook throwing exception on submission?"
            }
          ]
        },
        {
          "id": 2130547,
          "postDate": "2023-02-05T14:19:26.627Z",
          "content": "<p>Thanks.  I had similar problem.  I installed pylibjpeg_libjpeg instead of pylibjpeg.  My submission went through once with pylibjpeg_libjpeg.  But, always had submission error with others.  Hope this resolves the issue.</p>",
          "rawMarkdown": "Thanks.  I had similar problem.  I installed pylibjpeg_libjpeg instead of pylibjpeg.  My submission went through once with pylibjpeg_libjpeg.  But, always had submission error with others.  Hope this resolves the issue."
        }
      ]
    },
    {
      "id": 2058506,
      "postDate": "2022-12-08T01:44:44.563Z",
      "content": "<p>I use this <a href=\"https://www.kaggle.com/wqx2022/smbcd-pytorch-inference-sub\" target=\"_blank\">inference template</a> to generate submission.csv, everything seems normal. But my sub always Threw Exception, I kept my Internet OFF, I processed the DICOM to PNG into /kaggle/temp/XXX like someone's do, I did not use GPUs… It almost drives me mad, could anyone help me plz?</p>",
      "rawMarkdown": "I use this [inference template](https://www.kaggle.com/wqx2022/smbcd-pytorch-inference-sub) to generate submission.csv, everything seems normal. But my sub always Threw Exception, I kept my Internet OFF, I processed the DICOM to PNG into /kaggle/temp/XXX like someone's do, I did not use GPUs... It almost drives me mad, could anyone help me plz?"
    }
  ],
  "comments": [
    {
      "id": 2058575,
      "author_name": "Radek Osmulski",
      "author_url": "",
      "post_date": "2022-12-08T03:25:56.727000",
      "content": "<p>hey <a href=\"https://www.kaggle.com/wqx2022\" target=\"_blank\">@wqx2022</a>!</p>\n<p>There is a chance that missing this is the reason:</p>\n<pre><code>try:\n    import pylibjpeg\nexcept:\n    !rm -rf /root/.cache/torch/hub/checkpoints/\n    !mkdir -p /root/.cache/torch/hub/checkpoints/\n    !pip install /kaggle/input/rsna-2022-whl/{pydicom-2.3.0-py3-none-any.whl,pylibjpeg-1.4.0-py3-none-any.whl,python_gdcm-3.0.15-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl}\n    !pip install /kaggle/input/rsna-2022-whl/{torch-1.12.1-cp37-cp37m-manylinux1_x86_64.whl,torchvision-0.13.1-cp37-cp37m-manylinux1_x86_64.whl}\n</code></pre>\n<p>Without this you will be able to process the files in the visible test set, but not the hidden test set that gets attached to your VM later on.</p>\n<p>You also need to attach the <code>rsna-2022-whl</code> dataset to your notebook.</p>\n<p>BTW a good heuristic to try to identify what might be the issue is trying to predict on the train set. It has more files coming from different machines -- if your submission will run on some part of the train set, it will quite likely run on the hidden test set as well!</p>",
      "votes": 4,
      "replies": [
        {
          "id": 2058581,
          "author_name": "wqx2022",
          "author_url": "",
          "post_date": "2022-12-08T03:46:09.817000",
          "content": "<p>Yeah, thx, I'll try it later.</p>",
          "votes": 0,
          "replies": [
            {
              "id": 2125651,
              "author_name": "RohanHBTU",
              "author_url": "",
              "post_date": "2023-02-01T19:14:35.593000",
              "content": "<p>how did you solve your problem of notebook throwing exception on submission?</p>",
              "votes": 0,
              "replies": []
            }
          ]
        },
        {
          "id": 2130547,
          "author_name": "joejeo1",
          "author_url": "",
          "post_date": "2023-02-05T14:19:26.627000",
          "content": "<p>Thanks.  I had similar problem.  I installed pylibjpeg_libjpeg instead of pylibjpeg.  My submission went through once with pylibjpeg_libjpeg.  But, always had submission error with others.  Hope this resolves the issue.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2058575": "hey @wqx2022!\n\nThere is a chance that missing this is the reason:\n\n```\ntry:\n    import pylibjpeg\nexcept:\n    !rm -rf /root/.cache/torch/hub/checkpoints/\n    !mkdir -p /root/.cache/torch/hub/checkpoints/\n    !pip install /kaggle/input/rsna-2022-whl/{pydicom-2.3.0-py3-none-any.whl,pylibjpeg-1.4.0-py3-none-any.whl,python_gdcm-3.0.15-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl}\n    !pip install /kaggle/input/rsna-2022-whl/{torch-1.12.1-cp37-cp37m-manylinux1_x86_64.whl,torchvision-0.13.1-cp37-cp37m-manylinux1_x86_64.whl}\n```\nWithout this you will be able to process the files in the visible test set, but not the hidden test set that gets attached to your VM later on.\n\nYou also need to attach the `rsna-2022-whl` dataset to your notebook.\n\nBTW a good heuristic to try to identify what might be the issue is trying to predict on the train set. It has more files coming from different machines -- if your submission will run on some part of the train set, it will quite likely run on the hidden test set as well!",
    "2058506": "I use this [inference template](https://www.kaggle.com/wqx2022/smbcd-pytorch-inference-sub) to generate submission.csv, everything seems normal. But my sub always Threw Exception, I kept my Internet OFF, I processed the DICOM to PNG into /kaggle/temp/XXX like someone's do, I did not use GPUs... It almost drives me mad, could anyone help me plz?"
  }
}