{ "id": 254089, "title": "Questions about 9 hours run-time restrictions", "url": "/competitions/siim-covid19-detection/discussion/254089", "author_name": "", "post_date": "2021-07-20T04:25:35.917669500Z", "votes": null, "comment_count": 3, "views": 0, "content": "

Hi, I'm kaggle begginer.

\n

Usually in my code, it take about 5 hours to calculate when submitting.
\nBut according to leaderboard page, public leaderboard is calculated with approximately 29% of the test data while private leaderboard is 71%, and there is 9 hours run-time restriction for this competition.

\n

So I have some questions;
\n1) Is this competition's 9 hours run-time restriction harder than other comperitions?
\n2) Private leaderboard is calculated with about twice data than public, so calculating private leaderboards takes twice as long as public leaderboards?
\n3) Is there any way to reduce run-time when submitting?

", "messages": [ { "id": "1393974", "postDate": "07/20/2021 04:25:35", "content": "

Hi, I'm kaggle begginer.

\n

Usually in my code, it take about 5 hours to calculate when submitting.
\nBut according to leaderboard page, public leaderboard is calculated with approximately 29% of the test data while private leaderboard is 71%, and there is 9 hours run-time restriction for this competition.

\n

So I have some questions;
\n1) Is this competition's 9 hours run-time restriction harder than other comperitions?
\n2) Private leaderboard is calculated with about twice data than public, so calculating private leaderboards takes twice as long as public leaderboards?
\n3) Is there any way to reduce run-time when submitting?

", "rawMarkdown": "Hi, I'm kaggle begginer.\n\nUsually in my code, it take about 5 hours to calculate when submitting.\nBut according to leaderboard page, public leaderboard is calculated with approximately 29% of the test data while private leaderboard is 71%, and there is 9 hours run-time restriction for this competition.\n\nSo I have some questions;\n1) Is this competition's 9 hours run-time restriction harder than other comperitions?\n2) Private leaderboard is calculated with about twice data than public, so calculating private leaderboards takes twice as long as public leaderboards?\n3) Is there any way to reduce run-time when submitting?", "votes": null }, { "id": "1394133", "postDate": "07/20/2021 07:39:44", "content": "
    \n
  1. Better than stacking hundreds of submission csv files.
    2. \n
  2. When you make a submission, it is scored on both (public + private data), you're just shown public lb score. So, if a submission is successful and scored, need not panic about the time limit.
    3. \n
  3. There are many ways. removing unnecessary codes, higher test batch size, etc.
    4. \n
", "rawMarkdown": "1. Better than stacking hundreds of submission csv files.\n2. When you make a submission, it is scored on both (public + private data), you're just shown public lb score. So, if a submission is successful and scored, need not panic about the time limit.\n3. There are many ways. removing unnecessary codes, higher test batch size, etc.", "votes": null }, { "id": "1394362", "postDate": "07/20/2021 10:13:39", "content": "

thank you very much!

", "rawMarkdown": "thank you very much!", "votes": null }, { "id": "1394717", "postDate": "07/20/2021 14:47:31", "content": "

Twice data doesn't mean twice time. But you can try parallel methods to enhence your speed, such as allowing one thread to input data, another thread to control preprocessing.

", "rawMarkdown": "Twice data doesn't mean twice time. But you can try parallel methods to enhence your speed, such as allowing one thread to input data, another thread to control preprocessing.", "votes": null } ], "comments": [ { "id": 1394133, "author_name": "nischaydnk", "author_url": "", "post_date": "07/20/2021 07:39:44", "content": "
    \n
  1. Better than stacking hundreds of submission csv files.
    2. \n
  2. When you make a submission, it is scored on both (public + private data), you're just shown public lb score. So, if a submission is successful and scored, need not panic about the time limit.
    3. \n
  3. There are many ways. removing unnecessary codes, higher test batch size, etc.
    4. \n
", "votes": null, "replies": [ { "id": 1394362, "author_name": "welshonionman", "author_url": "", "post_date": "07/20/2021 10:13:39", "content": "

thank you very much!

", "votes": null, "replies": [] } ] }, { "id": 1394717, "author_name": "winniechavez", "author_url": "", "post_date": "07/20/2021 14:47:31", "content": "

Twice data doesn't mean twice time. But you can try parallel methods to enhence your speed, such as allowing one thread to input data, another thread to control preprocessing.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1393974": "Hi, I'm kaggle begginer.\n\nUsually in my code, it take about 5 hours to calculate when submitting.\nBut according to leaderboard page, public leaderboard is calculated with approximately 29% of the test data while private leaderboard is 71%, and there is 9 hours run-time restriction for this competition.\n\nSo I have some questions;\n1) Is this competition's 9 hours run-time restriction harder than other comperitions?\n2) Private leaderboard is calculated with about twice data than public, so calculating private leaderboards takes twice as long as public leaderboards?\n3) Is there any way to reduce run-time when submitting?", "1394133": "1. Better than stacking hundreds of submission csv files.\n2. When you make a submission, it is scored on both (public + private data), you're just shown public lb score. So, if a submission is successful and scored, need not panic about the time limit.\n3. There are many ways. removing unnecessary codes, higher test batch size, etc.", "1394362": "thank you very much!", "1394717": "Twice data doesn't mean twice time. But you can try parallel methods to enhence your speed, such as allowing one thread to input data, another thread to control preprocessing." }, "source": "meta" }