When I am saving the version and committing it takes 3 hours to generate a submission.csv file. But in the LB scoring, the code runs for more than 9 hours, and eventually, I got a time-out error.
\nIs there any way to solve this issue? How I can estimate my code running time on a private test dataset?
", "messages": [ { "id": "1481467", "postDate": "08/19/2021 14:19:34", "content": "When I am saving the version and committing it takes 3 hours to generate a submission.csv file. But in the LB scoring, the code runs for more than 9 hours, and eventually, I got a time-out error.
\nIs there any way to solve this issue? How I can estimate my code running time on a private test dataset?
", "rawMarkdown": "When I am saving the version and committing it takes 3 hours to generate a submission.csv file. But in the LB scoring, the code runs for more than 9 hours, and eventually, I got a time-out error.\n\nIs there any way to solve this issue? How I can estimate my code running time on a private test dataset?", "votes": null }, { "id": "1481698", "postDate": "08/19/2021 16:24:21", "content": "From the data description page:
\nNOTE: the total size of the rerun test set (Public and Private) is ~5x the size of the Public test set
I guess if your pipeline is 3h for the test folder then it should take 5 times more to run it on the whole data. If your code got a score without an error then you can be assured that there will be no problem on the private part
", "rawMarkdown": "From the data description page:\n`NOTE: the total size of the rerun test set (Public and Private) is ~5x the size of the Public test set`\n\nI guess if your pipeline is 3h for the test folder then it should take 5 times more to run it on the whole data. If your code got a score without an error then you can be assured that there will be no problem on the private part", "votes": null }, { "id": "1481727", "postDate": "08/19/2021 16:38:36", "content": "Thanks for your reply. Keeping that in mind I made modifications in the code to generate submission file within 1 hour 2 min and again submitted for public scoring, let's see whether I get time out error this time.
\nso if it takes 5 times more, then it should take around 5-6 hours for scoring a public score,
\nI am waiting to see what happens because while I was submitting DL models to generate the submission file it took 5 hours and the same amount of time for public scoring, that's why I am confused.
Also if you want just to see your public score, you can use fast submission like here
\nhttps://www.kaggle.com/narek1110/fast-submission-rsna
", "rawMarkdown": "Also if you want just to see your public score, you can use fast submission like here\n\nhttps://www.kaggle.com/narek1110/fast-submission-rsna", "votes": null }, { "id": "1481783", "postDate": "08/19/2021 17:00:55", "content": "I looked over it, that means I can test by only submitting my submission.csv file, right?
\nI mean to get an idea of whether my code is fine or not?
Yeah, in case if you want to validate the results of your experiment on the public leaderboard. That submission will not be valid as the final submission.
", "rawMarkdown": "Yeah, in case if you want to validate the results of your experiment on the public leaderboard. That submission will not be valid as the final submission.", "votes": null } ], "comments": [ { "id": 1481698, "author_name": "narek1110", "author_url": "", "post_date": "08/19/2021 16:24:21", "content": "From the data description page:
\nNOTE: the total size of the rerun test set (Public and Private) is ~5x the size of the Public test set
I guess if your pipeline is 3h for the test folder then it should take 5 times more to run it on the whole data. If your code got a score without an error then you can be assured that there will be no problem on the private part
", "votes": null, "replies": [ { "id": 1481727, "author_name": "wfarzana", "author_url": "", "post_date": "08/19/2021 16:38:36", "content": "Thanks for your reply. Keeping that in mind I made modifications in the code to generate submission file within 1 hour 2 min and again submitted for public scoring, let's see whether I get time out error this time.
\nso if it takes 5 times more, then it should take around 5-6 hours for scoring a public score,
\nI am waiting to see what happens because while I was submitting DL models to generate the submission file it took 5 hours and the same amount of time for public scoring, that's why I am confused.
Also if you want just to see your public score, you can use fast submission like here
\nhttps://www.kaggle.com/narek1110/fast-submission-rsna
", "votes": null, "replies": [] }, { "id": 1481783, "author_name": "wfarzana", "author_url": "", "post_date": "08/19/2021 17:00:55", "content": "I looked over it, that means I can test by only submitting my submission.csv file, right?
\nI mean to get an idea of whether my code is fine or not?
Yeah, in case if you want to validate the results of your experiment on the public leaderboard. That submission will not be valid as the final submission.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1481467": "When I am saving the version and committing it takes 3 hours to generate a submission.csv file. But in the LB scoring, the code runs for more than 9 hours, and eventually, I got a time-out error.\n\nIs there any way to solve this issue? How I can estimate my code running time on a private test dataset?", "1481698": "From the data description page:\n`NOTE: the total size of the rerun test set (Public and Private) is ~5x the size of the Public test set`\n\nI guess if your pipeline is 3h for the test folder then it should take 5 times more to run it on the whole data. If your code got a score without an error then you can be assured that there will be no problem on the private part", "1481727": "Thanks for your reply. Keeping that in mind I made modifications in the code to generate submission file within 1 hour 2 min and again submitted for public scoring, let's see whether I get time out error this time.\nso if it takes 5 times more, then it should take around 5-6 hours for scoring a public score, \nI am waiting to see what happens because while I was submitting DL models to generate the submission file it took 5 hours and the same amount of time for public scoring, that's why I am confused.", "1481747": "Also if you want just to see your public score, you can use fast submission like here\n\nhttps://www.kaggle.com/narek1110/fast-submission-rsna", "1481783": "I looked over it, that means I can test by only submitting my submission.csv file, right? \nI mean to get an idea of whether my code is fine or not?", "1481805": "Yeah, in case if you want to validate the results of your experiment on the public leaderboard. That submission will not be valid as the final submission." }, "source": "meta" }