{
  "id": 416183,
  "title": "How to solve “Notebook Threw Exception” in submission ??",
  "url": "/competitions/vesuvius-challenge-ink-detection/discussion/416183",
  "author_name": "",
  "post_date": "2023-06-10T03:13:50.311823700Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>My notebook based on 3D Resnet baseline [inference] ran successfully but encountered “Notebook Threw Exception” in submission. With denoising submission failed, Without denoising submission succeeded. Does anyone know how to solve this? </p>",
  "messages": [
    {
      "id": "2294387",
      "postDate": "06/10/2023 03:13:50",
      "content": "<p>My notebook based on 3D Resnet baseline [inference] ran successfully but encountered “Notebook Threw Exception” in submission. With denoising submission failed, Without denoising submission succeeded. Does anyone know how to solve this? </p>",
      "rawMarkdown": "My notebook based on 3D Resnet baseline [inference] ran successfully but encountered “Notebook Threw Exception” in submission. With denoising submission failed, Without denoising submission succeeded. Does anyone know how to solve this?",
      "votes": null
    },
    {
      "id": "2294543",
      "postDate": "06/10/2023 06:30:30",
      "content": "<p>I recall denoising code would cause a crash if I tried to use numpy in place of cupy (couldn't install on my kernel); perhaps package cupy not readily available in the competition kernel (use try-except to detect exact cause of exception thrown?); may be find  alternative implementations for denoising?</p>",
      "rawMarkdown": "I recall denoising code would cause a crash if I tried to use numpy in place of cupy (couldn't install on my kernel); perhaps package cupy not readily available in the competition kernel (use try-except to detect exact cause of exception thrown?); may be find  alternative implementations for denoising?",
      "votes": null
    },
    {
      "id": "2294845",
      "postDate": "06/10/2023 11:34:16",
      "content": "<p>You can use something like this to check where is the problem. </p>\n<pre><code> ():\n    submission = defaultdict()\n    submission[] = [, ]\n    submission[] = [, ]\n    pd.DataFrame.from_dict(submission).to_csv(, index=)\n     \n:\n    ...\n:\n   error_handle()\n</code></pre>",
      "rawMarkdown": "You can use something like this to check where is the problem. \n```\ndef error_handle():\n    submission = defaultdict(list)\n    submission['Id'] = ['a', 'b']\n    submission['Predicted'] = ['1 1 5 1', '10 20']\n    pd.DataFrame.from_dict(submission).to_csv(\"/kaggle/working/submission.csv\", index=False)\n    return None\ntry:\n    ...\nexcept:\n   error_handle()\n```",
      "votes": null
    },
    {
      "id": "2294876",
      "postDate": "06/10/2023 12:05:30",
      "content": "<p>i think (i tried it a t one point) its done on gpu so it could be vram issue just do a fast inference test with less stride (if its based on the baseline), batch size etc, or do on cpu …</p>\n<p>edit: cjheck the end comment: <a href=\"url\" target=\"_blank\">https://www.kaggle.com/code/yoyobar/3d-resnet-baseline-inference/comments</a></p>",
      "rawMarkdown": "i think (i tried it a t one point) its done on gpu so it could be vram issue just do a fast inference test with less stride (if its based on the baseline), batch size etc, or do on cpu ...\n\nedit: cjheck the end comment: [https://www.kaggle.com/code/yoyobar/3d-resnet-baseline-inference/comments](url)",
      "votes": null
    },
    {
      "id": "2294953",
      "postDate": "06/10/2023 13:35:03",
      "content": "<p>Thank you for sharing. I think this may be true for me. </p>",
      "rawMarkdown": "Thank you for sharing. I think this may be true for me.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2294543,
      "author_name": "vcolliym",
      "author_url": "",
      "post_date": "06/10/2023 06:30:30",
      "content": "<p>I recall denoising code would cause a crash if I tried to use numpy in place of cupy (couldn't install on my kernel); perhaps package cupy not readily available in the competition kernel (use try-except to detect exact cause of exception thrown?); may be find  alternative implementations for denoising?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2294845,
      "author_name": "jimmyisme1",
      "author_url": "",
      "post_date": "06/10/2023 11:34:16",
      "content": "<p>You can use something like this to check where is the problem. </p>\n<pre><code> ():\n    submission = defaultdict()\n    submission[] = [, ]\n    submission[] = [, ]\n    pd.DataFrame.from_dict(submission).to_csv(, index=)\n     \n:\n    ...\n:\n   error_handle()\n</code></pre>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2294876,
      "author_name": "iraqbot",
      "author_url": "",
      "post_date": "06/10/2023 12:05:30",
      "content": "<p>i think (i tried it a t one point) its done on gpu so it could be vram issue just do a fast inference test with less stride (if its based on the baseline), batch size etc, or do on cpu …</p>\n<p>edit: cjheck the end comment: <a href=\"url\" target=\"_blank\">https://www.kaggle.com/code/yoyobar/3d-resnet-baseline-inference/comments</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 2294953,
          "author_name": "clearwaterkzk",
          "author_url": "",
          "post_date": "06/10/2023 13:35:03",
          "content": "<p>Thank you for sharing. I think this may be true for me. </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2294387": "My notebook based on 3D Resnet baseline [inference] ran successfully but encountered “Notebook Threw Exception” in submission. With denoising submission failed, Without denoising submission succeeded. Does anyone know how to solve this?",
    "2294543": "I recall denoising code would cause a crash if I tried to use numpy in place of cupy (couldn't install on my kernel); perhaps package cupy not readily available in the competition kernel (use try-except to detect exact cause of exception thrown?); may be find  alternative implementations for denoising?",
    "2294845": "You can use something like this to check where is the problem. \n```\ndef error_handle():\n    submission = defaultdict(list)\n    submission['Id'] = ['a', 'b']\n    submission['Predicted'] = ['1 1 5 1', '10 20']\n    pd.DataFrame.from_dict(submission).to_csv(\"/kaggle/working/submission.csv\", index=False)\n    return None\ntry:\n    ...\nexcept:\n   error_handle()\n```",
    "2294876": "i think (i tried it a t one point) its done on gpu so it could be vram issue just do a fast inference test with less stride (if its based on the baseline), batch size etc, or do on cpu ...\n\nedit: cjheck the end comment: [https://www.kaggle.com/code/yoyobar/3d-resnet-baseline-inference/comments](url)",
    "2294953": "Thank you for sharing. I think this may be true for me."
  },
  "source": "meta"
}