Notebook Threw Exception Error- I got this error when i have submitted this - notebook . It has the entire pipeline from custom dataset creation for the model and final inference results saved to submission.csv
Submission Scoring Error - I got this error when i tried to submit just the output results from the previous notebook in the new notebook.
\nI tried sorting the rows as per the sample_submission.csv, yet i got this new error.
Can someone please look into this and help me out?
", "messages": [ { "id": "2978615", "postDate": "09/04/2024 05:42:26", "content": "Notebook Threw Exception Error- I got this error when i have submitted this - notebook . It has the entire pipeline from custom dataset creation for the model and final inference results saved to submission.csv
Submission Scoring Error - I got this error when i tried to submit just the output results from the previous notebook in the new notebook.
\nI tried sorting the rows as per the sample_submission.csv, yet i got this new error.
Can someone please look into this and help me out?
", "rawMarkdown": "1. Notebook Threw Exception Error- I got this error when i have submitted this - [notebook ](https://www.kaggle.com/code/rahulnakka/submissionrsna2024/edit/run/193540357). It has the entire pipeline from custom dataset creation for the model and final inference results saved to submission.csv\n\n2. Submission Scoring Error - I got this error when i tried to submit just the output results from the previous notebook in the new [notebook](https://www.kaggle.com/code/rahulnakka/notebook5046f7a019/edit). \nI tried sorting the rows as per the sample_submission.csv, yet i got this new error.\n\nCan someone please look into this and help me out?", "votes": null }, { "id": "2978677", "postDate": "09/04/2024 06:50:32", "content": "Execute it on train. to se what happens. If runs on train but submission fails, can be OOM.
", "rawMarkdown": "Execute it on train. to se what happens. If runs on train but submission fails, can be OOM.", "votes": null }, { "id": "2978690", "postDate": "09/04/2024 06:59:05", "content": "I tried the first notebook with train dataset and it worked, but failed on submission.
\nIn the second notebook, I just tried to submit the CSV file but got scoring error.
", "rawMarkdown": "I tried the first notebook with train dataset and it worked, but failed on submission. \n\nIn the second notebook, I just tried to submit the CSV file but got scoring error.", "votes": null }, { "id": "2978704", "postDate": "09/04/2024 07:08:00", "content": "Sure. It's a hidden test. You've only been able to submit local test, with a single sample. The code has to run over a hidden test with I assume more than one sample.
\nAnyway, don't waste daily submissions to debug. Run it locally over a chunk of train, over full train if you can. Many errors will pop in that scenario. If don't, my expirience say that can be one not notified OOM, so you can reduce batch sizes and reduce memmory usage.
\nSorry, but I've just woke up and I'm still sleepy, I'll check the code carefully when I can. But after a very very fast and superficial check looks very clean, so probably will be something easy to solve.
\nEDIT: I still need some sugar but I've see that you take sample from your own dataset, that contains a single sample. Take it from kaggle dataset, that will swap with real hidden test samples.
\nEDIT 2: Comment over the wrong notebook. Back to the correct one, do you save in memmory all crops that you will use? May be is OOM. Consider to generate crops ont he fly.
\nEDIT 3: Another recurrent sneaky error is not to clean working folder before save submission.csv
\nOften that causes submission error.
", "rawMarkdown": "Sure. It's a hidden test. You've only been able to submit local test, with a single sample. The code has to run over a hidden test with I assume more than one sample.\n\nAnyway, don't waste daily submissions to debug. Run it locally over a chunk of train, over full train if you can. Many errors will pop in that scenario. If don't, my expirience say that can be one not notified OOM, so you can reduce batch sizes and reduce memmory usage.\n\nSorry, but I've just woke up and I'm still sleepy, I'll check the code carefully when I can. But after a very very fast and superficial check looks very clean, so probably will be something easy to solve.\n\nEDIT: I still need some sugar but I've see that you take sample from your own dataset, that contains a single sample. Take it from kaggle dataset, that will swap with real hidden test samples.\n\nEDIT 2: Comment over the wrong notebook. Back to the correct one, do you save in memmory all crops that you will use? May be is OOM. Consider to generate crops ont he fly.\n\nEDIT 3: Another recurrent sneaky error is not to clean working folder before save submission.csv\n\nOften that causes submission error.", "votes": null }, { "id": "2980739", "postDate": "09/06/2024 06:09:18", "content": "try sub_df = sub_df.fillna(0.33) before saving the submission file.
\nIt worked for me.
Hi @rahulnakka ,
\nplease check this debug dataset if you are still facing issues with submission.
\n", "rawMarkdown": "Hi @rahulnakka ,\n\nplease check this debug dataset if you are still facing issues with submission.\n\n[debug dataset discussion](https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/523247)", "votes": null } ], "comments": [ { "id": 2978677, "author_name": "sacuscreed", "author_url": "", "post_date": "09/04/2024 06:50:32", "content": "Execute it on train. to se what happens. If runs on train but submission fails, can be OOM.
", "votes": null, "replies": [ { "id": 2978690, "author_name": "rahulnakka", "author_url": "", "post_date": "09/04/2024 06:59:05", "content": "I tried the first notebook with train dataset and it worked, but failed on submission.
\nIn the second notebook, I just tried to submit the CSV file but got scoring error.
", "votes": null, "replies": [ { "id": 2978704, "author_name": "sacuscreed", "author_url": "", "post_date": "09/04/2024 07:08:00", "content": "Sure. It's a hidden test. You've only been able to submit local test, with a single sample. The code has to run over a hidden test with I assume more than one sample.
\nAnyway, don't waste daily submissions to debug. Run it locally over a chunk of train, over full train if you can. Many errors will pop in that scenario. If don't, my expirience say that can be one not notified OOM, so you can reduce batch sizes and reduce memmory usage.
\nSorry, but I've just woke up and I'm still sleepy, I'll check the code carefully when I can. But after a very very fast and superficial check looks very clean, so probably will be something easy to solve.
\nEDIT: I still need some sugar but I've see that you take sample from your own dataset, that contains a single sample. Take it from kaggle dataset, that will swap with real hidden test samples.
\nEDIT 2: Comment over the wrong notebook. Back to the correct one, do you save in memmory all crops that you will use? May be is OOM. Consider to generate crops ont he fly.
\nEDIT 3: Another recurrent sneaky error is not to clean working folder before save submission.csv
\nOften that causes submission error.
", "votes": null, "replies": [] } ] } ] }, { "id": 2980739, "author_name": "rohitchaudhari25", "author_url": "", "post_date": "09/06/2024 06:09:18", "content": "try sub_df = sub_df.fillna(0.33) before saving the submission file.
\nIt worked for me.
Hi @rahulnakka ,
\nplease check this debug dataset if you are still facing issues with submission.
\n", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2978615": "1. Notebook Threw Exception Error- I got this error when i have submitted this - [notebook ](https://www.kaggle.com/code/rahulnakka/submissionrsna2024/edit/run/193540357). It has the entire pipeline from custom dataset creation for the model and final inference results saved to submission.csv\n\n2. Submission Scoring Error - I got this error when i tried to submit just the output results from the previous notebook in the new [notebook](https://www.kaggle.com/code/rahulnakka/notebook5046f7a019/edit). \nI tried sorting the rows as per the sample_submission.csv, yet i got this new error.\n\nCan someone please look into this and help me out?", "2978677": "Execute it on train. to se what happens. If runs on train but submission fails, can be OOM.", "2978690": "I tried the first notebook with train dataset and it worked, but failed on submission. \n\nIn the second notebook, I just tried to submit the CSV file but got scoring error.", "2978704": "Sure. It's a hidden test. You've only been able to submit local test, with a single sample. The code has to run over a hidden test with I assume more than one sample.\n\nAnyway, don't waste daily submissions to debug. Run it locally over a chunk of train, over full train if you can. Many errors will pop in that scenario. If don't, my expirience say that can be one not notified OOM, so you can reduce batch sizes and reduce memmory usage.\n\nSorry, but I've just woke up and I'm still sleepy, I'll check the code carefully when I can. But after a very very fast and superficial check looks very clean, so probably will be something easy to solve.\n\nEDIT: I still need some sugar but I've see that you take sample from your own dataset, that contains a single sample. Take it from kaggle dataset, that will swap with real hidden test samples.\n\nEDIT 2: Comment over the wrong notebook. Back to the correct one, do you save in memmory all crops that you will use? May be is OOM. Consider to generate crops ont he fly.\n\nEDIT 3: Another recurrent sneaky error is not to clean working folder before save submission.csv\n\nOften that causes submission error.", "2980739": "try `sub_df = sub_df.fillna(0.33)` before saving the submission file.\nIt worked for me.", "2992909": "Hi @rahulnakka ,\n\nplease check this debug dataset if you are still facing issues with submission.\n\n[debug dataset discussion](https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/523247)" }, "source": "meta" }