As we can see, the LB is flooding cause someone public a relative high scored notebook with all public dataset. It is so unfair to other people, our ranks could be dropped dramatically in one night(I dropped nearly 50 ranks).
\nAnd I am not saying it is wrong to run the code, I ran this code also, after I found the reason why my ranks dropped last night.
\nAgain, i am not against sharing, but we need a fair competition also, especially it is nearly the end of the competition.
\nI never encounter a circumstance like this, to be honest, I think the rank below that score is already meaningless, @wcukierski is there any solutions?
As we can see, the LB is flooding cause someone public a relative high scored notebook with all public dataset. It is so unfair to other people, our ranks could be dropped dramatically in one night(I dropped nearly 50 ranks).
\nAnd I am not saying it is wrong to run the code, I ran this code also, after I found the reason why my ranks dropped last night.
\nAgain, i am not against sharing, but we need a fair competition also, especially it is nearly the end of the competition.
\nI never encounter a circumstance like this, to be honest, I think the rank below that score is already meaningless, @wcukierski is there any solutions?
I think you have to deal with it, @kylewu . Recently, competitions I joined always have somebody shared their very high score kernel when it was about few days left. I think kaggle needs a way to solve this problem instead of just keep silence. Learning is a journey and not everyone is good at this field and on top of the LB. It's unfair for people who invest lots of time into competition but instead only got defeated by a public kernel. And one more thing is this LB is not the final LB, so keep fighting to the end.
", "rawMarkdown": "I think you have to deal with it, @kylewu . Recently, competitions I joined always have somebody shared their very high score kernel when it was about few days left. I think kaggle needs a way to solve this problem instead of just keep silence. Learning is a journey and not everyone is good at this field and on top of the LB. It's unfair for people who invest lots of time into competition but instead only got defeated by a public kernel. And one more thing is this LB is not the final LB, so keep fighting to the end.", "votes": null }, { "id": "1882035", "postDate": "08/03/2022 03:22:33", "content": "Let's consider the most extreme example: Kaggle's rules allow for the #1 scoring solution to be published on the last day of the competition.
", "rawMarkdown": "Let's consider the most extreme example: Kaggle's rules allow for the #1 scoring solution to be published on the last day of the competition.", "votes": null }, { "id": "1882444", "postDate": "08/03/2022 09:24:34", "content": "They must be banned forever :)
", "rawMarkdown": "They must be banned forever :)", "votes": null }, { "id": "1882643", "postDate": "08/03/2022 11:12:37", "content": "I think we need to update kaggle rules on this, or prohibit publishing notebooks in the last 2 weeks at the level of competition mechanics
", "rawMarkdown": "I think we need to update kaggle rules on this, or prohibit publishing notebooks in the last 2 weeks at the level of competition mechanics", "votes": null }, { "id": "1882751", "postDate": "08/03/2022 12:03:25", "content": "Thanks, man, will keep fighting and best luck to u.
", "rawMarkdown": "Thanks, man, will keep fighting and best luck to u.", "votes": null }, { "id": "1883076", "postDate": "08/03/2022 15:21:49", "content": "When there is a high scoring public notebook, you can chose to use it or not. If you use it and combine with your ideas then you should be able to improve over it.
", "rawMarkdown": "When there is a high scoring public notebook, you can chose to use it or not. If you use it and combine with your ideas then you should be able to improve over it.", "votes": null }, { "id": "1883090", "postDate": "08/03/2022 15:32:03", "content": "But not at the end of the competition.
", "rawMarkdown": "But not at the end of the competition.", "votes": null }, { "id": "1883220", "postDate": "08/03/2022 17:02:10", "content": "are we at the end of competition? I hope not.
", "rawMarkdown": "are we at the end of competition? I hope not.", "votes": null }, { "id": "1883373", "postDate": "08/03/2022 19:09:26", "content": "I experienced the same thing. And I understand how you feel. But is it a bad thing to publish better scores?
\nThere are some who are not very skilled in the field but are interested in the competition. Publishing the notebooks would help them to learn a lot (and faster in some cases). The best way to show that your solution is effective and relevant to the competition is to submit the result for a score. Sorry if this score is higher than someone else's score.
\nThe good news is that this solution made public is beneficial for everyone, including seniors (because we never finish learning).
\nAs @cpmpml said : \" if you use it and combine with your ideas then you should be able to improve over it \". That's the only solution for this problem (if indeed it is a problem).
Courage to us and good luck for the continuation.
", "rawMarkdown": "I experienced the same thing. And I understand how you feel. But is it a bad thing to publish better scores?\n\nThere are some who are not very skilled in the field but are interested in the competition. Publishing the notebooks would help them to learn a lot (and faster in some cases). The best way to show that your solution is effective and relevant to the competition is to submit the result for a score. Sorry if this score is higher than someone else's score.\n\nThe good news is that this solution made public is beneficial for everyone, including seniors (because we never finish learning). \nAs @cpmpml said : ` \" if you use it and combine with your ideas then you should be able to improve over it \"`. That's the only solution for this problem (if indeed it is a problem).\n\nCourage to us and good luck for the continuation.", "votes": null }, { "id": "1883486", "postDate": "08/03/2022 21:20:40", "content": "I understand very well how you feel.
\nBecause there was a time in the past when I stayed away from kaggle competitions because of those frustrations I had experienced several times.
\nHowever, I also think the decision to use it should be made by the user, not the platform. This is because it is difficult to distinguish between notebooks that are beneficial and notebooks that cause confusion without attrition.
\nThe leaderboards are a very important part of kaggle and hard to ignore, but I hope that you can find some element of enjoyment by looking not only at the leaderboards but also at your interest in the competition theme and the growth you have achieved before and after the competition.
\nWe are mostly \"unpaid labor\" so let's look for fun things to do.
", "rawMarkdown": "I understand very well how you feel. \n\nBecause there was a time in the past when I stayed away from kaggle competitions because of those frustrations I had experienced several times.\n\nHowever, I also think the decision to use it should be made by the user, not the platform. This is because it is difficult to distinguish between notebooks that are beneficial and notebooks that cause confusion without attrition.\n\nThe leaderboards are a very important part of kaggle and hard to ignore, but I hope that you can find some element of enjoyment by looking not only at the leaderboards but also at your interest in the competition theme and the growth you have achieved before and after the competition.\n\nWe are mostly \"unpaid labor\" so let's look for fun things to do.", "votes": null }, { "id": "1884624", "postDate": "08/04/2022 14:30:59", "content": "I know I'll be hopefully trying new ideas out through Sunday… whether or not they'll amount to anything is a good question ;)
\nThis is (a) the first time I've really done a competition in a while and (b) the first time I've used Transfomer deep learning, so this is much more a learning experience than about running up the leaderboard. (Getting to a silver medal would be nice, of course!)
\nAlso the notebook's poster has deleted data, so I think the notebook and derivatives are void for the private LB anyway for everyone who didn't run through the effort of copying out those model weights.
", "rawMarkdown": "I know I'll be hopefully trying new ideas out through Sunday... whether or not they'll amount to anything is a good question ;)\n\nThis is (a) the first time I've really done a competition in a while and (b) the first time I've used Transfomer deep learning, so this is much more a learning experience than about running up the leaderboard. (Getting to a silver medal would be nice, of course!)\n\nAlso the notebook's poster has deleted data, so I think the notebook and derivatives are void for the private LB anyway for everyone who didn't run through the effort of copying out those model weights.", "votes": null }, { "id": "1884775", "postDate": "08/04/2022 16:42:34", "content": "Notebook sharing is disabled in the final week of competitions - see Notebook Sharing Quiet Periods
", "rawMarkdown": "Notebook sharing is disabled in the final week of competitions - see [Notebook Sharing Quiet Periods](https://www.kaggle.com/general/291540)", "votes": null } ], "comments": [ { "id": 1882030, "author_name": "locbaop", "author_url": "", "post_date": "08/03/2022 03:13:34", "content": "I think you have to deal with it, @kylewu . Recently, competitions I joined always have somebody shared their very high score kernel when it was about few days left. I think kaggle needs a way to solve this problem instead of just keep silence. Learning is a journey and not everyone is good at this field and on top of the LB. It's unfair for people who invest lots of time into competition but instead only got defeated by a public kernel. And one more thing is this LB is not the final LB, so keep fighting to the end.
", "votes": null, "replies": [ { "id": 1882751, "author_name": "kylewu", "author_url": "", "post_date": "08/03/2022 12:03:25", "content": "Thanks, man, will keep fighting and best luck to u.
", "votes": null, "replies": [] } ] }, { "id": 1882035, "author_name": "cabbage972", "author_url": "", "post_date": "08/03/2022 03:22:33", "content": "Let's consider the most extreme example: Kaggle's rules allow for the #1 scoring solution to be published on the last day of the competition.
", "votes": null, "replies": [ { "id": 1884775, "author_name": "wcukierski", "author_url": "", "post_date": "08/04/2022 16:42:34", "content": "Notebook sharing is disabled in the final week of competitions - see Notebook Sharing Quiet Periods
", "votes": null, "replies": [] } ] }, { "id": 1882444, "author_name": "vad13irt", "author_url": "", "post_date": "08/03/2022 09:24:34", "content": "They must be banned forever :)
", "votes": null, "replies": [] }, { "id": 1882643, "author_name": "e0xextazy", "author_url": "", "post_date": "08/03/2022 11:12:37", "content": "I think we need to update kaggle rules on this, or prohibit publishing notebooks in the last 2 weeks at the level of competition mechanics
", "votes": null, "replies": [] }, { "id": 1883076, "author_name": "cpmpml", "author_url": "", "post_date": "08/03/2022 15:21:49", "content": "When there is a high scoring public notebook, you can chose to use it or not. If you use it and combine with your ideas then you should be able to improve over it.
", "votes": null, "replies": [ { "id": 1883090, "author_name": "e0xextazy", "author_url": "", "post_date": "08/03/2022 15:32:03", "content": "But not at the end of the competition.
", "votes": null, "replies": [] }, { "id": 1883220, "author_name": "cpmpml", "author_url": "", "post_date": "08/03/2022 17:02:10", "content": "are we at the end of competition? I hope not.
", "votes": null, "replies": [] }, { "id": 1884624, "author_name": "happycube", "author_url": "", "post_date": "08/04/2022 14:30:59", "content": "I know I'll be hopefully trying new ideas out through Sunday… whether or not they'll amount to anything is a good question ;)
\nThis is (a) the first time I've really done a competition in a while and (b) the first time I've used Transfomer deep learning, so this is much more a learning experience than about running up the leaderboard. (Getting to a silver medal would be nice, of course!)
\nAlso the notebook's poster has deleted data, so I think the notebook and derivatives are void for the private LB anyway for everyone who didn't run through the effort of copying out those model weights.
", "votes": null, "replies": [] } ] }, { "id": 1883373, "author_name": "cbeaud", "author_url": "", "post_date": "08/03/2022 19:09:26", "content": "I experienced the same thing. And I understand how you feel. But is it a bad thing to publish better scores?
\nThere are some who are not very skilled in the field but are interested in the competition. Publishing the notebooks would help them to learn a lot (and faster in some cases). The best way to show that your solution is effective and relevant to the competition is to submit the result for a score. Sorry if this score is higher than someone else's score.
\nThe good news is that this solution made public is beneficial for everyone, including seniors (because we never finish learning).
\nAs @cpmpml said : \" if you use it and combine with your ideas then you should be able to improve over it \". That's the only solution for this problem (if indeed it is a problem).
Courage to us and good luck for the continuation.
", "votes": null, "replies": [] }, { "id": 1883486, "author_name": "blue0620", "author_url": "", "post_date": "08/03/2022 21:20:40", "content": "I understand very well how you feel.
\nBecause there was a time in the past when I stayed away from kaggle competitions because of those frustrations I had experienced several times.
\nHowever, I also think the decision to use it should be made by the user, not the platform. This is because it is difficult to distinguish between notebooks that are beneficial and notebooks that cause confusion without attrition.
\nThe leaderboards are a very important part of kaggle and hard to ignore, but I hope that you can find some element of enjoyment by looking not only at the leaderboards but also at your interest in the competition theme and the growth you have achieved before and after the competition.
\nWe are mostly \"unpaid labor\" so let's look for fun things to do.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1881992": "As we can see, the LB is flooding cause someone public a relative high scored notebook with all public dataset. It is so unfair to other people, our ranks could be dropped dramatically in one night(I dropped nearly 50 ranks). \nAnd I am not saying it is wrong to run the code, I ran this code also, after I found the reason why my ranks dropped last night.\nAgain, i am not against sharing, but we need a fair competition also, especially it is nearly the end of the competition. \nI never encounter a circumstance like this, to be honest, I think the rank below that score is already meaningless, @wcukierski is there any solutions?", "1882030": "I think you have to deal with it, @kylewu . Recently, competitions I joined always have somebody shared their very high score kernel when it was about few days left. I think kaggle needs a way to solve this problem instead of just keep silence. Learning is a journey and not everyone is good at this field and on top of the LB. It's unfair for people who invest lots of time into competition but instead only got defeated by a public kernel. And one more thing is this LB is not the final LB, so keep fighting to the end.", "1882035": "Let's consider the most extreme example: Kaggle's rules allow for the #1 scoring solution to be published on the last day of the competition.", "1882444": "They must be banned forever :)", "1882643": "I think we need to update kaggle rules on this, or prohibit publishing notebooks in the last 2 weeks at the level of competition mechanics", "1882751": "Thanks, man, will keep fighting and best luck to u.", "1883076": "When there is a high scoring public notebook, you can chose to use it or not. If you use it and combine with your ideas then you should be able to improve over it.", "1883090": "But not at the end of the competition.", "1883220": "are we at the end of competition? I hope not.", "1883373": "I experienced the same thing. And I understand how you feel. But is it a bad thing to publish better scores?\n\nThere are some who are not very skilled in the field but are interested in the competition. Publishing the notebooks would help them to learn a lot (and faster in some cases). The best way to show that your solution is effective and relevant to the competition is to submit the result for a score. Sorry if this score is higher than someone else's score.\n\nThe good news is that this solution made public is beneficial for everyone, including seniors (because we never finish learning). \nAs @cpmpml said : ` \" if you use it and combine with your ideas then you should be able to improve over it \"`. That's the only solution for this problem (if indeed it is a problem).\n\nCourage to us and good luck for the continuation.", "1883486": "I understand very well how you feel. \n\nBecause there was a time in the past when I stayed away from kaggle competitions because of those frustrations I had experienced several times.\n\nHowever, I also think the decision to use it should be made by the user, not the platform. This is because it is difficult to distinguish between notebooks that are beneficial and notebooks that cause confusion without attrition.\n\nThe leaderboards are a very important part of kaggle and hard to ignore, but I hope that you can find some element of enjoyment by looking not only at the leaderboards but also at your interest in the competition theme and the growth you have achieved before and after the competition.\n\nWe are mostly \"unpaid labor\" so let's look for fun things to do.", "1884624": "I know I'll be hopefully trying new ideas out through Sunday... whether or not they'll amount to anything is a good question ;)\n\nThis is (a) the first time I've really done a competition in a while and (b) the first time I've used Transfomer deep learning, so this is much more a learning experience than about running up the leaderboard. (Getting to a silver medal would be nice, of course!)\n\nAlso the notebook's poster has deleted data, so I think the notebook and derivatives are void for the private LB anyway for everyone who didn't run through the effort of copying out those model weights.", "1884775": "Notebook sharing is disabled in the final week of competitions - see [Notebook Sharing Quiet Periods](https://www.kaggle.com/general/291540)" }, "source": "meta" }