{ "id": 437504, "title": "Notebook Submission Failed - Notebook Threw Exception", "url": "/competitions/rsna-2023-abdominal-trauma-detection/discussion/437504", "author_name": "wj66x419", "post_date": "2023-09-07T03:26:12.603000", "votes": 1, "comment_count": 5, "views": 0, "content": "

my notebook as below link
\nmy_notebbok

\n

I am a DL starter, it is my first time attending vision related competition on Kaggle.
\nMore than 5 times I have debugged my code, it still shows 'Notebook Threw Exception' after I submitted. The script works in interactive mode.

\n\n

I don't know what further I can do lol, Help me that any pointers on what could be the issue.

", "messages": [ { "id": 2475013, "postDate": "2023-10-09T15:32:35.847Z", "content": "

This problem may due to a corrupt image in the hidden test set when loading the pixel array with PIL.Image.
\nSkip that file in inferring may solve this exception.
\nDetailed information: https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/435815

", "rawMarkdown": "This problem may due to a corrupt image in the hidden test set when loading the pixel array with PIL.Image.\nSkip that file in inferring may solve this exception.\nDetailed information: https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/435815\n ", "votes": 1 }, { "id": 2430003, "postDate": "2023-09-09T01:42:09.757Z", "content": "

Just some ideas as I also had problem submitting.

\n

You may need to make prediction on each patient one by one, because the hidden test sample are much larger than those 3 patients in the sample submission.

\n

Another way to debug is try running your evaluation on the training data to debug. You can calculate your score check out this notebook. https://www.kaggle.com/code/metric/rsna-trauma-metric/notebook

\n

I wrote my own notebook. It's a bit messy but hopefully its helpful https://www.kaggle.com/code/franklinshih0617/cnn-training-with-png

", "rawMarkdown": "Just some ideas as I also had problem submitting. \n\nYou may need to make prediction on each patient one by one, because the hidden test sample are much larger than those 3 patients in the sample submission.\n\nAnother way to debug is try running your evaluation on the training data to debug. You can calculate your score check out this notebook. https://www.kaggle.com/code/metric/rsna-trauma-metric/notebook\n\nI wrote my own notebook. It's a bit messy but hopefully its helpful https://www.kaggle.com/code/franklinshih0617/cnn-training-with-png", "votes": 1, "replies": [ { "id": 2430120, "postDate": "2023-09-09T04:51:55.060Z", "content": "

Hey! Could you make your notebook public, I am also going through a submission error roadblock.

", "rawMarkdown": "Hey! Could you make your notebook public, I am also going through a submission error roadblock.", "replies": [ { "id": 2432591, "postDate": "2023-09-11T02:48:36.183Z", "content": "

I just made it public! Hopefully it helps!

", "rawMarkdown": "I just made it public! Hopefully it helps!" } ] } ] }, { "id": 2427006, "postDate": "2023-09-07T03:26:12.603Z", "content": "

my notebook as below link
\nmy_notebbok

\n

I am a DL starter, it is my first time attending vision related competition on Kaggle.
\nMore than 5 times I have debugged my code, it still shows 'Notebook Threw Exception' after I submitted. The script works in interactive mode.

\n\n

I don't know what further I can do lol, Help me that any pointers on what could be the issue.

", "rawMarkdown": "my notebook as below link\n[my_notebbok](https://www.kaggle.com/wj66x419/pytorch-cnn-baseline-model)\n\nI am a DL starter, it is my first time attending vision related competition on Kaggle.\nMore than 5 times I have debugged my code, it still shows 'Notebook Threw Exception' after I submitted. The script works in interactive mode.\n* filled NA \n* standard Dicoms[Standardizing Unusual Dicoms](https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/427217)\n* group by patient\n* float_format='%.20f'\n\nI don't know what further I can do lol, Help me that any pointers on what could be the issue.", "votes": 1 }, { "id": 2428440, "postDate": "2023-09-07T21:56:13.117Z", "content": "

I'm experiencing a similar problem, but with the submissions per day limit it's painful to debug. In my case, exception doesn't seem to be related to the hidden test data, so probably occurs somewhere in my train data preprocessing cells or my model training cells.

\n

It would be useful if the organisers could give us a clue about what's different between the interactive NB environment and the one used to evaluate our scripts, or else ensure that when a script runs correctly in interactive mode it also runs correctly when evaluating.

", "rawMarkdown": "I'm experiencing a similar problem, but with the submissions per day limit it's painful to debug. In my case, exception doesn't seem to be related to the hidden test data, so probably occurs somewhere in my train data preprocessing cells or my model training cells. \n\nIt would be useful if the organisers could give us a clue about what's different between the interactive NB environment and the one used to evaluate our scripts, or else ensure that when a script runs correctly in interactive mode it also runs correctly when evaluating. ", "votes": 2 } ], "comments": [ { "id": 2475013, "author_name": "bighandsome2022", "author_url": "", "post_date": "2023-10-09T15:32:35.847000", "content": "

This problem may due to a corrupt image in the hidden test set when loading the pixel array with PIL.Image.
\nSkip that file in inferring may solve this exception.
\nDetailed information: https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/435815

", "votes": 1, "replies": [] }, { "id": 2430003, "author_name": "Franklin Shih0617", "author_url": "", "post_date": "2023-09-09T01:42:09.757000", "content": "

Just some ideas as I also had problem submitting.

\n

You may need to make prediction on each patient one by one, because the hidden test sample are much larger than those 3 patients in the sample submission.

\n

Another way to debug is try running your evaluation on the training data to debug. You can calculate your score check out this notebook. https://www.kaggle.com/code/metric/rsna-trauma-metric/notebook

\n

I wrote my own notebook. It's a bit messy but hopefully its helpful https://www.kaggle.com/code/franklinshih0617/cnn-training-with-png

", "votes": 1, "replies": [ { "id": 2430120, "author_name": "gabriel rudloff", "author_url": "", "post_date": "2023-09-09T04:51:55.060000", "content": "

Hey! Could you make your notebook public, I am also going through a submission error roadblock.

", "votes": 0, "replies": [ { "id": 2432591, "author_name": "Franklin Shih0617", "author_url": "", "post_date": "2023-09-11T02:48:36.183000", "content": "

I just made it public! Hopefully it helps!

", "votes": 0, "replies": [] } ] } ] }, { "id": 2428440, "author_name": "A Humphrey", "author_url": "", "post_date": "2023-09-07T21:56:13.117000", "content": "

I'm experiencing a similar problem, but with the submissions per day limit it's painful to debug. In my case, exception doesn't seem to be related to the hidden test data, so probably occurs somewhere in my train data preprocessing cells or my model training cells.

\n

It would be useful if the organisers could give us a clue about what's different between the interactive NB environment and the one used to evaluate our scripts, or else ensure that when a script runs correctly in interactive mode it also runs correctly when evaluating.

", "votes": 2, "replies": [] } ], "raw_markdown_by_id": { "2475013": "This problem may due to a corrupt image in the hidden test set when loading the pixel array with PIL.Image.\nSkip that file in inferring may solve this exception.\nDetailed information: https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/435815\n ", "2430003": "Just some ideas as I also had problem submitting. \n\nYou may need to make prediction on each patient one by one, because the hidden test sample are much larger than those 3 patients in the sample submission.\n\nAnother way to debug is try running your evaluation on the training data to debug. You can calculate your score check out this notebook. https://www.kaggle.com/code/metric/rsna-trauma-metric/notebook\n\nI wrote my own notebook. It's a bit messy but hopefully its helpful https://www.kaggle.com/code/franklinshih0617/cnn-training-with-png", "2427006": "my notebook as below link\n[my_notebbok](https://www.kaggle.com/wj66x419/pytorch-cnn-baseline-model)\n\nI am a DL starter, it is my first time attending vision related competition on Kaggle.\nMore than 5 times I have debugged my code, it still shows 'Notebook Threw Exception' after I submitted. The script works in interactive mode.\n* filled NA \n* standard Dicoms[Standardizing Unusual Dicoms](https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/discussion/427217)\n* group by patient\n* float_format='%.20f'\n\nI don't know what further I can do lol, Help me that any pointers on what could be the issue.", "2428440": "I'm experiencing a similar problem, but with the submissions per day limit it's painful to debug. In my case, exception doesn't seem to be related to the hidden test data, so probably occurs somewhere in my train data preprocessing cells or my model training cells. \n\nIt would be useful if the organisers could give us a clue about what's different between the interactive NB environment and the one used to evaluate our scripts, or else ensure that when a script runs correctly in interactive mode it also runs correctly when evaluating. " } }