I identified an issue on our end that was causing some notebooks to fail improperly. Fixing it and publishing the scores looks likely to be done by the end of the business day PST tomorrow.
", "messages": [ { "id": "1693657", "postDate": "02/16/2022 20:56:49", "content": "I identified an issue on our end that was causing some notebooks to fail improperly. Fixing it and publishing the scores looks likely to be done by the end of the business day PST tomorrow.
", "rawMarkdown": "I identified an issue on our end that was causing some notebooks to fail improperly. Fixing it and publishing the scores looks likely to be done by the end of the business day PST tomorrow.", "votes": null }, { "id": "1693662", "postDate": "02/16/2022 21:03:27", "content": "Thanks for the update with the re-run 👍
", "rawMarkdown": "Thanks for the update with the re-run 👍", "votes": null }, { "id": "1693688", "postDate": "02/16/2022 21:42:09", "content": "Any explanation for the change in public leaderboard scores and ranking?
", "rawMarkdown": "Any explanation for the change in public leaderboard scores and ranking?", "votes": null }, { "id": "1694409", "postDate": "02/17/2022 12:10:32", "content": "In 'more normal' competitions, we submit and both the Public and Private test set are run on that moment, so nothing can change later and we are good to go.
\nIn 'evaluate later' competitions, it is much more tricky.
\nWhy so? : people submitted based on architecture/code/assumptions (including assumptions related to data) of submission period. So any deviation from existing, 'might' potentially save some, but break others.
\nI know, as kaggle staff you know this more than anybody else and will handle this delicate issue accordingly.
\nMy final words on this: take your time. If it is not today, be it. I (or anybody) would rather get a late score, rather than getting a no score.
\nThanks for all your efforts.
", "rawMarkdown": "In 'more normal' competitions, we submit and both the Public and Private test set are run on that moment, so nothing can change later and we are good to go.\n\nIn 'evaluate later' competitions, it is much more tricky. \n\nWhy so? : people submitted based on architecture/code/assumptions (including assumptions related to data) of submission period. So any deviation from existing, 'might' potentially save some, but break others. \n\nI know, as kaggle staff you know this more than anybody else and will handle this delicate issue accordingly. \n\nMy final words on this: take your time. If it is not today, be it. I (or anybody) would rather get a late score, rather than getting a no score. \n\nThanks for all your efforts.", "votes": null }, { "id": "1694500", "postDate": "02/17/2022 13:26:38", "content": "Agreed - it's harder, but the upside is that like this the competition is truly free from leakage and is closer to real-life data science
", "rawMarkdown": "Agreed - it's harder, but the upside is that like this the competition is truly free from leakage and is closer to real-life data science", "votes": null }, { "id": "1694562", "postDate": "02/17/2022 14:20:23", "content": "Maybe https://time.is/PST had some extra hits yesterday/today ;)
\nBut quote @abdulkadirguner
\n\n", "rawMarkdown": "Maybe https://time.is/PST had some extra hits yesterday/today ;)\n\nBut quote @abdulkadirguner \n\n> take your time. If it is not today, be it. I (or anybody) would rather get a late score, rather than getting a no score.", "votes": null }, { "id": "1695321", "postDate": "02/18/2022 04:38:25", "content": "take your time. If it is not today, be it. I (or anybody) would rather get a late score, rather than getting a no score.
\n
Plus, There are also 3 months to go in the competition, major software face outages, unless its in the last week of the competition, we shouldn't trouble the Kaggle team too much :)
", "rawMarkdown": "Plus, There are also 3 months to go in the competition, major software face outages, unless its in the last week of the competition, we shouldn't trouble the Kaggle team too much :)", "votes": null } ], "comments": [ { "id": 1693662, "author_name": "kirderf", "author_url": "", "post_date": "02/16/2022 21:03:27", "content": "Thanks for the update with the re-run 👍
", "votes": null, "replies": [] }, { "id": 1693688, "author_name": "julianmukaj", "author_url": "", "post_date": "02/16/2022 21:42:09", "content": "Any explanation for the change in public leaderboard scores and ranking?
", "votes": null, "replies": [] }, { "id": 1694409, "author_name": "abdulkadirguner", "author_url": "", "post_date": "02/17/2022 12:10:32", "content": "In 'more normal' competitions, we submit and both the Public and Private test set are run on that moment, so nothing can change later and we are good to go.
\nIn 'evaluate later' competitions, it is much more tricky.
\nWhy so? : people submitted based on architecture/code/assumptions (including assumptions related to data) of submission period. So any deviation from existing, 'might' potentially save some, but break others.
\nI know, as kaggle staff you know this more than anybody else and will handle this delicate issue accordingly.
\nMy final words on this: take your time. If it is not today, be it. I (or anybody) would rather get a late score, rather than getting a no score.
\nThanks for all your efforts.
", "votes": null, "replies": [ { "id": 1694500, "author_name": "marcogorelli", "author_url": "", "post_date": "02/17/2022 13:26:38", "content": "Agreed - it's harder, but the upside is that like this the competition is truly free from leakage and is closer to real-life data science
", "votes": null, "replies": [] }, { "id": 1695321, "author_name": "init27", "author_url": "", "post_date": "02/18/2022 04:38:25", "content": "Plus, There are also 3 months to go in the competition, major software face outages, unless its in the last week of the competition, we shouldn't trouble the Kaggle team too much :)
", "votes": null, "replies": [] } ] }, { "id": 1694562, "author_name": "kirderf", "author_url": "", "post_date": "02/17/2022 14:20:23", "content": "Maybe https://time.is/PST had some extra hits yesterday/today ;)
\nBut quote @abdulkadirguner
\n\n", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1693657": "I identified an issue on our end that was causing some notebooks to fail improperly. Fixing it and publishing the scores looks likely to be done by the end of the business day PST tomorrow.", "1693662": "Thanks for the update with the re-run 👍", "1693688": "Any explanation for the change in public leaderboard scores and ranking?", "1694409": "In 'more normal' competitions, we submit and both the Public and Private test set are run on that moment, so nothing can change later and we are good to go.\n\nIn 'evaluate later' competitions, it is much more tricky. \n\nWhy so? : people submitted based on architecture/code/assumptions (including assumptions related to data) of submission period. So any deviation from existing, 'might' potentially save some, but break others. \n\nI know, as kaggle staff you know this more than anybody else and will handle this delicate issue accordingly. \n\nMy final words on this: take your time. If it is not today, be it. I (or anybody) would rather get a late score, rather than getting a no score. \n\nThanks for all your efforts.", "1694500": "Agreed - it's harder, but the upside is that like this the competition is truly free from leakage and is closer to real-life data science", "1694562": "Maybe https://time.is/PST had some extra hits yesterday/today ;)\n\nBut quote @abdulkadirguner \n\n> take your time. If it is not today, be it. I (or anybody) would rather get a late score, rather than getting a no score.", "1695321": "Plus, There are also 3 months to go in the competition, major software face outages, unless its in the last week of the competition, we shouldn't trouble the Kaggle team too much :)" }, "source": "meta" }take your time. If it is not today, be it. I (or anybody) would rather get a late score, rather than getting a no score.
\n