in the past few days, many people have started posting high scoring notebooks given only 5 days are left I don't think it is fair to people who worked hard in the competition
", "messages": [ { "id": 1471598, "postDate": "2021-08-14T10:21:41.320Z", "content": "in the past few days, many people have started posting high scoring notebooks given only 5 days are left I don't think it is fair to people who worked hard in the competition
", "rawMarkdown": "in the past few days, many people have started posting high scoring notebooks given only 5 days are left I don't think it is fair to people who worked hard in the competition", "votes": 36 }, { "id": 1474034, "postDate": "2021-08-15T21:14:43.680Z", "content": "Fully agree with this. Making and stating the rules is one thing … but some Kagglers will keep breaking those rules if they are not enforced publicly.
\n@inversion @maggiemd May'be a nice moment to take some action on this? Would be a nice signal to the rule breakers and will for sure be appreciated by the Kagglers who try to follow the rules.
", "rawMarkdown": "Fully agree with this. Making and stating the rules is one thing ... but some Kagglers will keep breaking those rules if they are not enforced publicly.\n\n@inversion @maggiemd May'be a nice moment to take some action on this? Would be a nice signal to the rule breakers and will for sure be appreciated by the Kagglers who try to follow the rules.", "votes": 9 }, { "id": 1473983, "postDate": "2021-08-15T20:12:34.807Z", "content": "There is a rule against it, but no1 cares. This is by far the most frustrating thing of Kaggle
\nhttps://www.kaggle.com/spoiler-alert
It is really against the morals of the competition and has many bad effects on other sides too.
\nit is literally said https://www.kaggle.com/spoiler-alert that posting high score notebooks is disruptive near the deadlines.
I think it is very unfortunate to see so many high scoring notebooks being published in the last week … and I feel sad to see many hard work being undermined
\nand people just copy and get higher scores… can beat weeks and in months of hard work…
We should \"Report notebook\" for these cases to highlight as @coldfir3 says its against the spirit of the competition
\nand hopefully Kaggle team will take action again this.
Either they are valuable, and you can benefit from them by blending with your own work, or they are worthless and you can ignore them.
\nEither way there is a positive spin out of this late sharing.
\nThis said, I agree with you that late sharing should be forbidden.
", "rawMarkdown": "Either they are valuable, and you can benefit from them by blending with your own work, or they are worthless and you can ignore them.\n\nEither way there is a positive spin out of this late sharing.\n\nThis said, I agree with you that late sharing should be forbidden.", "votes": 3, "replies": [ { "id": 1475760, "postDate": "2021-08-16T20:23:57.127Z", "rawMarkdown": "", "isDeleted": true }, { "id": 1476094, "postDate": "2021-08-17T01:49:25.513Z", "content": "@cpmpml with all due respect, blending does require using diverse models so if someone were already using a similar model/approach as a part of their solution they will not gain significant score boost whereas someone having diverse models can get ahead which I don't think is fair as they probably wouldn't have gotten the lead in the end if it weren't for the share
", "rawMarkdown": "@cpmpml with all due respect, blending does require using diverse models so if someone were already using a similar model/approach as a part of their solution they will not gain significant score boost whereas someone having diverse models can get ahead which I don't think is fair as they probably wouldn't have gotten the lead in the end if it weren't for the share", "votes": 3 }, { "id": 1476195, "postDate": "2021-08-17T03:11:47.217Z", "content": "I think you just proved CPMP's point though. Someone using stock public kernels with only slight tinkering won't get much millage (and realistically, probably doesn't even know how to stack/blend/ensemble for reals); whereas one who's invested in their own distinct pre/post processing strategies and modeling will only come out a winner here. That stated, after a certain gap, the public kernels just pull your ensemble scores down though…
", "rawMarkdown": "I think you just proved CPMP's point though. Someone using stock public kernels with only slight tinkering won't get much millage (and realistically, probably doesn't even know how to stack/blend/ensemble for reals); whereas one who's invested in their own distinct pre/post processing strategies and modeling will only come out a winner here. That stated, after a certain gap, the public kernels just pull your ensemble scores down though...", "votes": 2 }, { "id": 1476267, "postDate": "2021-08-17T04:10:23.797Z", "content": "@mrigendraagrawal - there will always be arguments for and against and have heard them all.
\nThe point raised consistently is if Kaggle takes a view it should not be done, but then still rewards these notebooks with medals - it is like saying \"you're grounded but have a chocolate bar on your way out and here's the car keys too and take some money from my wallet… but don't forget you are grounded!\"
It all becomes pointless and devalues the platform and maybe people give up that would not otherwise. All you can do really is learn what you can and want to here, achieve the goals you set for yourself and block out all the rest. Best of luck!
", "rawMarkdown": "@mrigendraagrawal - there will always be arguments for and against and have heard them all. \nThe point raised consistently is if Kaggle takes a view it should not be done, but then still rewards these notebooks with medals - it is like saying \"you're grounded but have a chocolate bar on your way out and here's the car keys too and take some money from my wallet... but don't forget you are grounded!\" \n\nIt all becomes pointless and devalues the platform and maybe people give up that would not otherwise. All you can do really is learn what you can and want to here, achieve the goals you set for yourself and block out all the rest. Best of luck! ", "votes": 1 }, { "id": 1476793, "postDate": "2021-08-17T08:50:06.747Z", "content": "\n\nEither they are valuable, and you can benefit from them by blending with your own work, …
\n
Often times they are not reproducible and if you strictly follow the rules you should not use them in your solution. But many still do, so you are in a dilemma.
", "rawMarkdown": "> Either they are valuable, and you can benefit from them by blending with your own work, ...\n\nOften times they are not reproducible and if you strictly follow the rules you should not use them in your solution. But many still do, so you are in a dilemma.", "votes": 6 } ] }, { "id": 1474399, "postDate": "2021-08-16T04:56:16.763Z", "content": "One thing that has seemed odd is the number of teams keeps increasing well past the team merge date. Guess since this competition had a restart maybe some have just come back at the end (?) but think when you have some high scoring notebooks it adds to this increasing number (as I posted elsewhere it would). Kind of would prefer if a competition restarts then you join again.
\nBut will have to see if the ensembles of ensembles hold positions or not when it finishes. Right now just want this to end!
Fully agree with this. Making and stating the rules is one thing … but some Kagglers will keep breaking those rules if they are not enforced publicly.
\n@inversion @maggiemd May'be a nice moment to take some action on this? Would be a nice signal to the rule breakers and will for sure be appreciated by the Kagglers who try to follow the rules.
", "votes": 9, "replies": [] }, { "id": 1473983, "author_name": "Adriano Passos", "author_url": "", "post_date": "2021-08-15T20:12:34.807000", "content": "There is a rule against it, but no1 cares. This is by far the most frustrating thing of Kaggle
\nhttps://www.kaggle.com/spoiler-alert
It is really against the morals of the competition and has many bad effects on other sides too.
\nit is literally said https://www.kaggle.com/spoiler-alert that posting high score notebooks is disruptive near the deadlines.
I think it is very unfortunate to see so many high scoring notebooks being published in the last week … and I feel sad to see many hard work being undermined
\nand people just copy and get higher scores… can beat weeks and in months of hard work…
We should \"Report notebook\" for these cases to highlight as @coldfir3 says its against the spirit of the competition
\nand hopefully Kaggle team will take action again this.
Either they are valuable, and you can benefit from them by blending with your own work, or they are worthless and you can ignore them.
\nEither way there is a positive spin out of this late sharing.
\nThis said, I agree with you that late sharing should be forbidden.
", "votes": 3, "replies": [ { "id": 1475760, "author_name": "", "author_url": "", "post_date": "2021-08-16T20:23:57.127000", "content": "", "votes": 0, "replies": [] }, { "id": 1476094, "author_name": "shiroe", "author_url": "", "post_date": "2021-08-17T01:49:25.513000", "content": "@cpmpml with all due respect, blending does require using diverse models so if someone were already using a similar model/approach as a part of their solution they will not gain significant score boost whereas someone having diverse models can get ahead which I don't think is fair as they probably wouldn't have gotten the lead in the end if it weren't for the share
", "votes": 3, "replies": [] }, { "id": 1476195, "author_name": "عثمان", "author_url": "", "post_date": "2021-08-17T03:11:47.217000", "content": "I think you just proved CPMP's point though. Someone using stock public kernels with only slight tinkering won't get much millage (and realistically, probably doesn't even know how to stack/blend/ensemble for reals); whereas one who's invested in their own distinct pre/post processing strategies and modeling will only come out a winner here. That stated, after a certain gap, the public kernels just pull your ensemble scores down though…
", "votes": 2, "replies": [] }, { "id": 1476267, "author_name": "something4kag", "author_url": "", "post_date": "2021-08-17T04:10:23.797000", "content": "@mrigendraagrawal - there will always be arguments for and against and have heard them all.
\nThe point raised consistently is if Kaggle takes a view it should not be done, but then still rewards these notebooks with medals - it is like saying \"you're grounded but have a chocolate bar on your way out and here's the car keys too and take some money from my wallet… but don't forget you are grounded!\"
It all becomes pointless and devalues the platform and maybe people give up that would not otherwise. All you can do really is learn what you can and want to here, achieve the goals you set for yourself and block out all the rest. Best of luck!
", "votes": 1, "replies": [] }, { "id": 1476793, "author_name": "Psi", "author_url": "", "post_date": "2021-08-17T08:50:06.747000", "content": "\n\nEither they are valuable, and you can benefit from them by blending with your own work, …
\n
Often times they are not reproducible and if you strictly follow the rules you should not use them in your solution. But many still do, so you are in a dilemma.
", "votes": 6, "replies": [] } ] }, { "id": 1474399, "author_name": "something4kag", "author_url": "", "post_date": "2021-08-16T04:56:16.763000", "content": "One thing that has seemed odd is the number of teams keeps increasing well past the team merge date. Guess since this competition had a restart maybe some have just come back at the end (?) but think when you have some high scoring notebooks it adds to this increasing number (as I posted elsewhere it would). Kind of would prefer if a competition restarts then you join again.
\nBut will have to see if the ensembles of ensembles hold positions or not when it finishes. Right now just want this to end!