TL;DR: There is a rule against it: https://www.kaggle.com/spoiler-alert
\nThis competition is one of the biggest examples of a deep problem Kaggle's system has. Sharing notebooks should be done in a way to share knowledge, not for simple farming upvotes and easy medals. By far the most frustrating thing about Kaggle is to work hard for a whole week to make a submission just to find out that 80% of the participants just copied and pasted a submission. That is not only unfair with the others but makes it near impossible to grasp your true progress.
\nThis happens over and over in all competitions but this specific one is absolutely bad. We have dozens of 'ensemble' notebooks that don't provide any useful insight or knowledge and yet they ruin the hard work of hundreds of people. Rules of Kaggle comps should change! there shouldn't be allowed to share results after a certain deadline of like a month or so before the comp ends.
", "messages": [ { "id": 1507728, "postDate": "2021-09-09T13:39:50.277Z", "content": "TL;DR: There is a rule against it: https://www.kaggle.com/spoiler-alert
\nThis competition is one of the biggest examples of a deep problem Kaggle's system has. Sharing notebooks should be done in a way to share knowledge, not for simple farming upvotes and easy medals. By far the most frustrating thing about Kaggle is to work hard for a whole week to make a submission just to find out that 80% of the participants just copied and pasted a submission. That is not only unfair with the others but makes it near impossible to grasp your true progress.
\nThis happens over and over in all competitions but this specific one is absolutely bad. We have dozens of 'ensemble' notebooks that don't provide any useful insight or knowledge and yet they ruin the hard work of hundreds of people. Rules of Kaggle comps should change! there shouldn't be allowed to share results after a certain deadline of like a month or so before the comp ends.
", "rawMarkdown": "TL;DR: There is a rule against it: https://www.kaggle.com/spoiler-alert\n\nThis competition is one of the biggest examples of a deep problem Kaggle's system has. Sharing notebooks should be done in a way to share knowledge, not for simple farming upvotes and easy medals. By far the most frustrating thing about Kaggle is to work hard for a whole week to make a submission just to find out that 80% of the participants just copied and pasted a submission. That is not only unfair with the others but makes it near impossible to grasp your true progress.\n\nThis happens over and over in all competitions but this specific one is absolutely bad. We have dozens of 'ensemble' notebooks that don't provide any useful insight or knowledge and yet they ruin the hard work of hundreds of people. Rules of Kaggle comps should change! there shouldn't be allowed to share results after a certain deadline of like a month or so before the comp ends.\n\n", "votes": 51 }, { "id": 1507842, "postDate": "2021-09-09T15:34:24.190Z", "content": "It is fine until last week. Posting 3 weeks before end allows people to leverage it before deadline.
\nIf public notebook beats you then by combining your work with the public notebook you should be in front of it.
", "rawMarkdown": "It is fine until last week. Posting 3 weeks before end allows people to leverage it before deadline.\n\nIf public notebook beats you then by combining your work with the public notebook you should be in front of it.", "votes": 21 }, { "id": 1508128, "postDate": "2021-09-09T22:11:23.463Z", "content": "This is a community trying collaboratively to solve a difficult problem spanning data science and physics (as well as competing for medals). Sharing of good codes allows others to \"stand on the shoulders of giants\" and improve the notebooks still further. If Kaggle competitions were about 1000 teams locked in their own silos and working independently, I don't think we would do such a good job of achieving excellent solutions. There are still nearly three weeks left; please feel free to keep sharing until the last week, when the warning banner goes up - at which time the disruption to competitive balance at so late a stage is considered to outweigh the benefit of seeding further ideas.
", "rawMarkdown": "This is a community trying collaboratively to solve a difficult problem spanning data science and physics (as well as competing for medals). Sharing of good codes allows others to \"stand on the shoulders of giants\" and improve the notebooks still further. If Kaggle competitions were about 1000 teams locked in their own silos and working independently, I don't think we would do such a good job of achieving excellent solutions. There are still nearly three weeks left; please feel free to keep sharing until the last week, when the warning banner goes up - at which time the disruption to competitive balance at so late a stage is considered to outweigh the benefit of seeding further ideas.", "votes": 18 }, { "id": 1515263, "postDate": "2021-09-16T22:37:09.737Z", "content": "For everyone saying Kaggle is hard:
\nPlease start posting OOFs with your high scoring notebooks ;-)
", "rawMarkdown": "Please start posting OOFs with your high scoring notebooks ;-)", "votes": 13, "replies": [ { "id": 1525759, "postDate": "2021-09-27T15:55:11.090Z", "content": "I'm pleasantly surprised no one did it o_O. I was certain that there'd be more higher scoring kernels as we got close to landing, but it appears that will not be the case.
", "rawMarkdown": "I'm pleasantly surprised no one did it o_O. I was certain that there'd be more higher scoring kernels as we got close to landing, but it appears that will not be the case.", "votes": 1 }, { "id": 1525775, "postDate": "2021-09-27T16:07:55.557Z", "content": "Considering the tight gap of scores on the leaderboard, I don't think anyone would want to ruin their own efforts for a notebook medal. 😄
", "rawMarkdown": "Considering the tight gap of scores on the leaderboard, I don't think anyone would want to ruin their own efforts for a notebook medal. 😄", "votes": 1 } ] }, { "id": 1507740, "postDate": "2021-09-09T13:53:34.977Z", "content": "Actually, we have 20 days (3 weeks) to go. It is still early.
\nAt the current development rate, the top public lb could be in the range of 0.887 in the end. gold may be around 0.883.
\nCurrent notebooks in the range of 0.87x are actually not very high.
", "rawMarkdown": "Actually, we have 20 days (3 weeks) to go. It is still early.\n\nAt the current development rate, the top public lb could be in the range of 0.887 in the end. gold may be around 0.883.\n\nCurrent notebooks in the range of 0.87x are actually not very high.", "votes": 14, "replies": [ { "id": 1528203, "postDate": "2021-09-29T13:11:55.757Z", "content": "\"At the current development rate, the top public lb could be in the range of 0.887 in the end. gold may be around 0.883.\"
\nand also the methods in the top teams are not coded in the notebook.
\nnow looking back, there is no need to worry in the first place for the then so-called high-scoring notebook ….
\nbcauase these are \"standard engineering solutions\" and nothing special.
maybe tmr when the solutions are revealed tmr, you will see that some of the solutions are in the forum and not in the code.
\nsome solutions are not even discussed and were \"unknown\" to many.
\nthe high-scoring notebook actually saves us time to show results if you trained with large network and large image size. this set baseline for the top solutions.
", "rawMarkdown": "\"At the current development rate, the top public lb could be in the range of 0.887 in the end. gold may be around 0.883.\"\n\nand also the methods in the top teams are not coded in the notebook.\n\n----\nnow looking back, there is no need to worry in the first place for the then so-called high-scoring notebook ....\nbcauase these are \"standard engineering solutions\" and nothing special.\n\nmaybe tmr when the solutions are revealed tmr, you will see that some of the solutions are in the forum and not in the code. \n\nsome solutions are not even discussed and were \"unknown\" to many.\n\nthe high-scoring notebook actually saves us time to show results if you trained with large network and large image size. this set baseline for the top solutions.\n", "votes": 3 } ] }, { "id": 1515617, "postDate": "2021-09-17T09:35:49.410Z", "content": "Measure your success on what you have learnt rather on how the public score looks like. There are always very interesting behaviour patterns in each competition. One of them is the \"ensemble swarm\" once a few high-scoring kernels appeared. And another one is also the shake-up. ;-) Don't trust the public LB, start focussing on your CV and your comparison with the public LB, try to balance!
\nThere is nothing bad on building ensembles as it has it's own goal to reduce overfitting. It's also not bad to built upon ideas that had been shared by the community. But instead of playing a \"point-and-click\" adventure of the random kind it's definitely better to built something you understand (at least to some point).
\nWithout learning something during a competition all Kaggle medals are just an illusion of competence that can be easily dissolved when someone starts asking knowledge related questions.
\nSo don't worry! :-)
", "rawMarkdown": "Measure your success on what you have learnt rather on how the public score looks like. There are always very interesting behaviour patterns in each competition. One of them is the \"ensemble swarm\" once a few high-scoring kernels appeared. And another one is also the shake-up. ;-) Don't trust the public LB, start focussing on your CV and your comparison with the public LB, try to balance! \n\nThere is nothing bad on building ensembles as it has it's own goal to reduce overfitting. It's also not bad to built upon ideas that had been shared by the community. But instead of playing a \"point-and-click\" adventure of the random kind it's definitely better to built something you understand (at least to some point). \n\nWithout learning something during a competition all Kaggle medals are just an illusion of competence that can be easily dissolved when someone starts asking knowledge related questions.\n\nSo don't worry! :-) ", "votes": 7, "replies": [ { "id": 1515635, "postDate": "2021-09-17T09:58:26.433Z", "content": "This 100%. The best way to measure progress is to stop comparing yourself to others and compare yourself to who you were yesterday. Medal counts mean very little.
\nMaybe I should have saved that inspirational quote for LinkedIn haha 😂
", "rawMarkdown": "This 100%. The best way to measure progress is to stop comparing yourself to others and compare yourself to who you were yesterday. Medal counts mean very little.\n\nMaybe I should have saved that inspirational quote for LinkedIn haha 😂", "votes": 12 } ] }, { "id": 1513849, "postDate": "2021-09-15T13:22:03.593Z", "content": "These notebooks do help people like me who are just starting out to learn and get better but yeah there should be like a right time to share these. Can't really see what the people who just copy their submissions are hoping to achieve. Like how long will you keep this up, the best will still beat you by the end and you really get nothing out of it.
", "rawMarkdown": "These notebooks do help people like me who are just starting out to learn and get better but yeah there should be like a right time to share these. Can't really see what the people who just copy their submissions are hoping to achieve. Like how long will you keep this up, the best will still beat you by the end and you really get nothing out of it.", "votes": 4 }, { "id": 1510497, "postDate": "2021-09-12T13:11:27.813Z", "content": "I lost my interest in Kaggle because of the ensemble kernels shared.
", "rawMarkdown": "I lost my interest in Kaggle because of the ensemble kernels shared.", "votes": 2, "replies": [ { "id": 1510890, "postDate": "2021-09-12T21:05:55.420Z", "content": "Kaggle starts to get really interesting when you learn enough to beat public ensembles.
\nIf you fail to do that in one competition, then read high-placed solutions on forums after competition ends. Implement them, tweak them, understand them, remember them. Next competiton would be significantly easier.
\nYou will rise above the public notebooks eventually.
", "rawMarkdown": "Kaggle starts to get really interesting when you learn enough to beat public ensembles.\n\nIf you fail to do that in one competition, then read high-placed solutions on forums after competition ends. Implement them, tweak them, understand them, remember them. Next competiton would be significantly easier.\n\nYou will rise above the public notebooks eventually.", "votes": 15 } ] }, { "id": 1514194, "postDate": "2021-09-15T19:02:47.960Z", "content": "This is why it is always hard to participate in comp as a newbie…
", "rawMarkdown": "This is why it is always hard to participate in comp as a newbie...", "votes": 1 }, { "id": 1513742, "postDate": "2021-09-15T11:33:25.177Z", "content": "High scoring notebooks should not be allowed in last 3-4 weeks.
", "rawMarkdown": "High scoring notebooks should not be allowed in last 3-4 weeks.", "votes": 1, "replies": [ { "id": 1514252, "postDate": "2021-09-15T20:42:23.503Z", "content": "and this should be enforced by the Kaggle system itself
", "rawMarkdown": "and this should be enforced by the Kaggle system itself", "votes": 3 }, { "id": 1515693, "postDate": "2021-09-17T11:43:55.620Z", "content": "Yes, Kaggle should enforce a rule that if for a notebook, LB score is better than 100th rank score and last 30 days are left, then its score should not be shown to users.
", "rawMarkdown": "Yes, Kaggle should enforce a rule that if for a notebook, LB score is better than 100th rank score and last 30 days are left, then its score should not be shown to users.", "votes": -1 } ] }, { "id": 1512379, "postDate": "2021-09-14T08:08:52.840Z", "content": "I found public notebooks (and specially those with high score) to be really informative about the good pratice in data science (here in deep learning). I don't think we should remove them because those are really informative and this is what make kaggle interesting.
\nBut as the author say it is really frustrating to see people just copied / paste public submission. It makes the competition frustrating. As other comment put it, we should just stop public submission that are too high in the last days of the competition (~2 weeks before the deadline).
", "rawMarkdown": "I found public notebooks (and specially those with high score) to be really informative about the good pratice in data science (here in deep learning). I don't think we should remove them because those are really informative and this is what make kaggle interesting.\n\nBut as the author say it is really frustrating to see people just copied / paste public submission. It makes the competition frustrating. As other comment put it, we should just stop public submission that are too high in the last days of the competition (~2 weeks before the deadline).", "votes": 1, "replies": [ { "id": 1512532, "postDate": "2021-09-14T10:35:10.977Z", "content": "You won't find many people getting medals solely by download-uploading or copy-pasting a public model. The cut-off for bronze will generally be some way above the best public notebook (in most designs of competition). Assuming that people comply with the moratorium in the last week, we have at least seven days to beat the top public submission.
\nUnless you are near the very top of the leaderboard, it's not likely that whatever score you've posted 16 days before the end (that's now as I write) will be sufficient for a medal, simply because the standard of models rises towards the end of a competition. So we should expect to have to work hard in the final week.
", "rawMarkdown": "You won't find many people getting medals solely by download-uploading or copy-pasting a public model. The cut-off for bronze will generally be some way above the best public notebook (in most designs of competition). Assuming that people comply with the moratorium in the last week, we have at least seven days to beat the top public submission. \n\nUnless you are near the very top of the leaderboard, it's not likely that whatever score you've posted 16 days before the end (that's now as I write) will be sufficient for a medal, simply because the standard of models rises towards the end of a competition. So we should expect to have to work hard in the final week.", "votes": 2 } ] }, { "id": 1507835, "postDate": "2021-09-09T15:28:03.500Z", "content": "while I agree these notebooks are stupid/annoying they do tell you a little about who is merging in public notebooks into their ensemble (you see some teams dramatically shift when a public-no-real-info notebook is released)
", "rawMarkdown": "while I agree these notebooks are stupid/annoying they do tell you a little about who is merging in public notebooks into their ensemble (you see some teams dramatically shift when a public-no-real-info notebook is released)", "votes": 1, "replies": [ { "id": 1507836, "postDate": "2021-09-09T15:30:04.030Z", "content": "So everyone, judging by current rate of leaderboard movement?:D
", "rawMarkdown": "So everyone, judging by current rate of leaderboard movement?:D", "votes": 4 }, { "id": 1507838, "postDate": "2021-09-09T15:31:13.997Z", "content": "Haha yeah exactly 🤣
", "rawMarkdown": "Haha yeah exactly 🤣" } ] }, { "id": 1511169, "postDate": "2021-09-13T07:16:41.893Z", "content": "Let us stand on the shoulders of giants😛
", "rawMarkdown": "Let us stand on the shoulders of giants😛", "votes": 2 }, { "id": 1509173, "postDate": "2021-09-11T01:26:36.153Z", "content": "Cant believe there are dummy people copying others code. You are learning nothing. And for that Kaggle is, to learn. if you want money get a f job.
", "rawMarkdown": "Cant believe there are dummy people copying others code. You are learning nothing. And for that Kaggle is, to learn. if you want money get a f****** job.", "votes": -2, "replies": [ { "id": 1512377, "postDate": "2021-09-14T08:07:18.347Z", "content": "Nice point. I guess you meant to say copying without playing with it in order to improve the ideas behind. Then I totally agree.
", "rawMarkdown": "Nice point. I guess you meant to say copying without playing with it in order to improve the ideas behind. Then I totally agree. ", "votes": 2 }, { "id": 1512695, "postDate": "2021-09-14T13:48:35.250Z", "content": "Yes ofc. If you improve what you copied means you totally understand others work and you can even make it better. But it is not the first time i see a post where someone complains that people sometimes is just copying and pasting other codes in order to be at top in competitions.
", "rawMarkdown": "Yes ofc. If you improve what you copied means you totally understand others work and you can even make it better. But it is not the first time i see a post where someone complains that people sometimes is just copying and pasting other codes in order to be at top in competitions.", "votes": 1 } ] }, { "id": 1515112, "postDate": "2021-09-16T18:07:45.397Z", "content": "Is there a dislike button on kaggle?)
", "rawMarkdown": "Is there a dislike button on kaggle?)", "votes": 1 }, { "id": 1510871, "postDate": "2021-09-12T20:23:40.750Z", "content": "According me there must not be any public notebook for active competition…. Discussion is okay… 🧐🧐
", "rawMarkdown": "According me there must not be any public notebook for active competition.... Discussion is okay... 🧐🧐", "votes": -4 }, { "id": 1559742, "postDate": "2021-10-27T07:11:00.530Z", "content": "Hey All,
\nThank you all for taking part in our competition. The participation has been overwhelmingly positive. We are currently conducting a survey to gauge the demographic and outreach achieved. Kindly spare 2min and fill in this survey https://forms.gle/QP9L16niPexozyhu5.
\nThank you all,
\nRegards,
\nChris
Kaggle is fun
", "rawMarkdown": "Kaggle is fun" }, { "id": 1515262, "postDate": "2021-09-16T22:32:30.650Z", "content": "Interesting
", "rawMarkdown": "Interesting" }, { "id": 1512518, "postDate": "2021-09-14T10:12:15.710Z", "content": "Hear Hear !
", "rawMarkdown": "Hear Hear !" }, { "id": 1511060, "postDate": "2021-09-13T04:49:48.283Z", "content": "I also find it really hard to participate as an individual
", "rawMarkdown": "I also find it really hard to participate as an individual", "isDeleted": true, "replies": [ { "id": 1512378, "postDate": "2021-09-14T08:08:45.840Z", "content": "I'm a starter here. What do you find especially hard while participating alone? Is it something Kaggle specific or generally it's hard to do as much experiments as a team can do?
", "rawMarkdown": "I'm a starter here. What do you find especially hard while participating alone? Is it something Kaggle specific or generally it's hard to do as much experiments as a team can do? " } ] } ], "comments": [ { "id": 1507842, "author_name": "CPMP", "author_url": "", "post_date": "2021-09-09T15:34:24.190000", "content": "It is fine until last week. Posting 3 weeks before end allows people to leverage it before deadline.
\nIf public notebook beats you then by combining your work with the public notebook you should be in front of it.
", "votes": 21, "replies": [] }, { "id": 1508128, "author_name": "John Mitchell", "author_url": "", "post_date": "2021-09-09T22:11:23.463000", "content": "This is a community trying collaboratively to solve a difficult problem spanning data science and physics (as well as competing for medals). Sharing of good codes allows others to \"stand on the shoulders of giants\" and improve the notebooks still further. If Kaggle competitions were about 1000 teams locked in their own silos and working independently, I don't think we would do such a good job of achieving excellent solutions. There are still nearly three weeks left; please feel free to keep sharing until the last week, when the warning banner goes up - at which time the disruption to competitive balance at so late a stage is considered to outweigh the benefit of seeding further ideas.
", "votes": 18, "replies": [] }, { "id": 1515263, "author_name": "datasaurus", "author_url": "", "post_date": "2021-09-16T22:37:09.737000", "content": "For everyone saying Kaggle is hard:
\nPlease start posting OOFs with your high scoring notebooks ;-)
", "votes": 13, "replies": [ { "id": 1525759, "author_name": "عثمان", "author_url": "", "post_date": "2021-09-27T15:55:11.090000", "content": "I'm pleasantly surprised no one did it o_O. I was certain that there'd be more higher scoring kernels as we got close to landing, but it appears that will not be the case.
", "votes": 1, "replies": [] }, { "id": 1525775, "author_name": "Nischay Dhankhar", "author_url": "", "post_date": "2021-09-27T16:07:55.557000", "content": "Considering the tight gap of scores on the leaderboard, I don't think anyone would want to ruin their own efforts for a notebook medal. 😄
", "votes": 1, "replies": [] } ] }, { "id": 1507740, "author_name": "hengck23", "author_url": "", "post_date": "2021-09-09T13:53:34.977000", "content": "Actually, we have 20 days (3 weeks) to go. It is still early.
\nAt the current development rate, the top public lb could be in the range of 0.887 in the end. gold may be around 0.883.
\nCurrent notebooks in the range of 0.87x are actually not very high.
", "votes": 14, "replies": [ { "id": 1528203, "author_name": "hengck23", "author_url": "", "post_date": "2021-09-29T13:11:55.757000", "content": "\"At the current development rate, the top public lb could be in the range of 0.887 in the end. gold may be around 0.883.\"
\nand also the methods in the top teams are not coded in the notebook.
\nnow looking back, there is no need to worry in the first place for the then so-called high-scoring notebook ….
\nbcauase these are \"standard engineering solutions\" and nothing special.
maybe tmr when the solutions are revealed tmr, you will see that some of the solutions are in the forum and not in the code.
\nsome solutions are not even discussed and were \"unknown\" to many.
\nthe high-scoring notebook actually saves us time to show results if you trained with large network and large image size. this set baseline for the top solutions.
", "votes": 3, "replies": [] } ] }, { "id": 1515617, "author_name": "Laura Fink", "author_url": "", "post_date": "2021-09-17T09:35:49.410000", "content": "Measure your success on what you have learnt rather on how the public score looks like. There are always very interesting behaviour patterns in each competition. One of them is the \"ensemble swarm\" once a few high-scoring kernels appeared. And another one is also the shake-up. ;-) Don't trust the public LB, start focussing on your CV and your comparison with the public LB, try to balance!
\nThere is nothing bad on building ensembles as it has it's own goal to reduce overfitting. It's also not bad to built upon ideas that had been shared by the community. But instead of playing a \"point-and-click\" adventure of the random kind it's definitely better to built something you understand (at least to some point).
\nWithout learning something during a competition all Kaggle medals are just an illusion of competence that can be easily dissolved when someone starts asking knowledge related questions.
\nSo don't worry! :-)
", "votes": 7, "replies": [ { "id": 1515635, "author_name": "datasaurus", "author_url": "", "post_date": "2021-09-17T09:58:26.433000", "content": "This 100%. The best way to measure progress is to stop comparing yourself to others and compare yourself to who you were yesterday. Medal counts mean very little.
\nMaybe I should have saved that inspirational quote for LinkedIn haha 😂
", "votes": 12, "replies": [] } ] }, { "id": 1513849, "author_name": "Harpreet Singh Wazir", "author_url": "", "post_date": "2021-09-15T13:22:03.593000", "content": "These notebooks do help people like me who are just starting out to learn and get better but yeah there should be like a right time to share these. Can't really see what the people who just copy their submissions are hoping to achieve. Like how long will you keep this up, the best will still beat you by the end and you really get nothing out of it.
", "votes": 4, "replies": [] }, { "id": 1510497, "author_name": "Tolga", "author_url": "", "post_date": "2021-09-12T13:11:27.813000", "content": "I lost my interest in Kaggle because of the ensemble kernels shared.
", "votes": 2, "replies": [ { "id": 1510890, "author_name": "fffrrt", "author_url": "", "post_date": "2021-09-12T21:05:55.420000", "content": "Kaggle starts to get really interesting when you learn enough to beat public ensembles.
\nIf you fail to do that in one competition, then read high-placed solutions on forums after competition ends. Implement them, tweak them, understand them, remember them. Next competiton would be significantly easier.
\nYou will rise above the public notebooks eventually.
", "votes": 15, "replies": [] } ] }, { "id": 1514194, "author_name": "Yash Gupta", "author_url": "", "post_date": "2021-09-15T19:02:47.960000", "content": "This is why it is always hard to participate in comp as a newbie…
", "votes": 1, "replies": [] }, { "id": 1513742, "author_name": "Anurag Trivedi", "author_url": "", "post_date": "2021-09-15T11:33:25.177000", "content": "High scoring notebooks should not be allowed in last 3-4 weeks.
", "votes": 1, "replies": [ { "id": 1514252, "author_name": "Adriano Passos", "author_url": "", "post_date": "2021-09-15T20:42:23.503000", "content": "and this should be enforced by the Kaggle system itself
", "votes": 3, "replies": [] }, { "id": 1515693, "author_name": "Anurag Trivedi", "author_url": "", "post_date": "2021-09-17T11:43:55.620000", "content": "Yes, Kaggle should enforce a rule that if for a notebook, LB score is better than 100th rank score and last 30 days are left, then its score should not be shown to users.
", "votes": -1, "replies": [] } ] }, { "id": 1512379, "author_name": "AdrienB", "author_url": "", "post_date": "2021-09-14T08:08:52.840000", "content": "I found public notebooks (and specially those with high score) to be really informative about the good pratice in data science (here in deep learning). I don't think we should remove them because those are really informative and this is what make kaggle interesting.
\nBut as the author say it is really frustrating to see people just copied / paste public submission. It makes the competition frustrating. As other comment put it, we should just stop public submission that are too high in the last days of the competition (~2 weeks before the deadline).
", "votes": 1, "replies": [ { "id": 1512532, "author_name": "John Mitchell", "author_url": "", "post_date": "2021-09-14T10:35:10.977000", "content": "You won't find many people getting medals solely by download-uploading or copy-pasting a public model. The cut-off for bronze will generally be some way above the best public notebook (in most designs of competition). Assuming that people comply with the moratorium in the last week, we have at least seven days to beat the top public submission.
\nUnless you are near the very top of the leaderboard, it's not likely that whatever score you've posted 16 days before the end (that's now as I write) will be sufficient for a medal, simply because the standard of models rises towards the end of a competition. So we should expect to have to work hard in the final week.
", "votes": 2, "replies": [] } ] }, { "id": 1507835, "author_name": "Hamish", "author_url": "", "post_date": "2021-09-09T15:28:03.500000", "content": "while I agree these notebooks are stupid/annoying they do tell you a little about who is merging in public notebooks into their ensemble (you see some teams dramatically shift when a public-no-real-info notebook is released)
", "votes": 1, "replies": [ { "id": 1507836, "author_name": "fffrrt", "author_url": "", "post_date": "2021-09-09T15:30:04.030000", "content": "So everyone, judging by current rate of leaderboard movement?:D
", "votes": 4, "replies": [] }, { "id": 1507838, "author_name": "Hamish", "author_url": "", "post_date": "2021-09-09T15:31:13.997000", "content": "Haha yeah exactly 🤣
", "votes": 0, "replies": [] } ] }, { "id": 1511169, "author_name": "Hinokami Kaggler", "author_url": "", "post_date": "2021-09-13T07:16:41.893000", "content": "Let us stand on the shoulders of giants😛
", "votes": 2, "replies": [] }, { "id": 1509173, "author_name": "Juan Jose Cesar Cabral", "author_url": "", "post_date": "2021-09-11T01:26:36.153000", "content": "Cant believe there are dummy people copying others code. You are learning nothing. And for that Kaggle is, to learn. if you want money get a f job.
", "votes": -2, "replies": [ { "id": 1512377, "author_name": "Yauhen", "author_url": "", "post_date": "2021-09-14T08:07:18.347000", "content": "Nice point. I guess you meant to say copying without playing with it in order to improve the ideas behind. Then I totally agree.
", "votes": 2, "replies": [] }, { "id": 1512695, "author_name": "Juan Jose Cesar Cabral", "author_url": "", "post_date": "2021-09-14T13:48:35.250000", "content": "Yes ofc. If you improve what you copied means you totally understand others work and you can even make it better. But it is not the first time i see a post where someone complains that people sometimes is just copying and pasting other codes in order to be at top in competitions.
", "votes": 1, "replies": [] } ] }, { "id": 1515112, "author_name": "remkamal", "author_url": "", "post_date": "2021-09-16T18:07:45.397000", "content": "Is there a dislike button on kaggle?)
", "votes": 1, "replies": [] }, { "id": 1510871, "author_name": "Yash Gupta", "author_url": "", "post_date": "2021-09-12T20:23:40.750000", "content": "According me there must not be any public notebook for active competition…. Discussion is okay… 🧐🧐
", "votes": -4, "replies": [] }, { "id": 1559742, "author_name": "ChristopherZerafa", "author_url": "", "post_date": "2021-10-27T07:11:00.530000", "content": "Hey All,
\nThank you all for taking part in our competition. The participation has been overwhelmingly positive. We are currently conducting a survey to gauge the demographic and outreach achieved. Kindly spare 2min and fill in this survey https://forms.gle/QP9L16niPexozyhu5.
\nThank you all,
\nRegards,
\nChris
Kaggle is fun
", "votes": 0, "replies": [] }, { "id": 1515262, "author_name": "Shuzhi Zeng", "author_url": "", "post_date": "2021-09-16T22:32:30.650000", "content": "Interesting
", "votes": 0, "replies": [] }, { "id": 1512518, "author_name": "Ippudkippude", "author_url": "", "post_date": "2021-09-14T10:12:15.710000", "content": "Hear Hear !
", "votes": 0, "replies": [] }, { "id": 1511060, "author_name": "", "author_url": "", "post_date": "2021-09-13T04:49:48.283000", "content": "I also find it really hard to participate as an individual
", "votes": 0, "replies": [ { "id": 1512378, "author_name": "Yauhen", "author_url": "", "post_date": "2021-09-14T08:08:45.840000", "content": "I'm a starter here. What do you find especially hard while participating alone? Is it something Kaggle specific or generally it's hard to do as much experiments as a team can do?
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "1507728": "TL;DR: There is a rule against it: https://www.kaggle.com/spoiler-alert\n\nThis competition is one of the biggest examples of a deep problem Kaggle's system has. Sharing notebooks should be done in a way to share knowledge, not for simple farming upvotes and easy medals. By far the most frustrating thing about Kaggle is to work hard for a whole week to make a submission just to find out that 80% of the participants just copied and pasted a submission. That is not only unfair with the others but makes it near impossible to grasp your true progress.\n\nThis happens over and over in all competitions but this specific one is absolutely bad. We have dozens of 'ensemble' notebooks that don't provide any useful insight or knowledge and yet they ruin the hard work of hundreds of people. Rules of Kaggle comps should change! there shouldn't be allowed to share results after a certain deadline of like a month or so before the comp ends.\n\n", "1507842": "It is fine until last week. Posting 3 weeks before end allows people to leverage it before deadline.\n\nIf public notebook beats you then by combining your work with the public notebook you should be in front of it.", "1508128": "This is a community trying collaboratively to solve a difficult problem spanning data science and physics (as well as competing for medals). Sharing of good codes allows others to \"stand on the shoulders of giants\" and improve the notebooks still further. If Kaggle competitions were about 1000 teams locked in their own silos and working independently, I don't think we would do such a good job of achieving excellent solutions. There are still nearly three weeks left; please feel free to keep sharing until the last week, when the warning banner goes up - at which time the disruption to competitive balance at so late a stage is considered to outweigh the benefit of seeding further ideas.", "1515263": "For everyone saying Kaggle is hard:\n1. Remember you are competing against the best in the world - don't be hard on yourself\n2. I'm sure everyone at the top will agree - it never gets easy! I find every competition hard. It wouldn't be fun if it was easy", "1508121": "Please start posting OOFs with your high scoring notebooks ;-)", "1507740": "Actually, we have 20 days (3 weeks) to go. It is still early.\n\nAt the current development rate, the top public lb could be in the range of 0.887 in the end. gold may be around 0.883.\n\nCurrent notebooks in the range of 0.87x are actually not very high.", "1515617": "Measure your success on what you have learnt rather on how the public score looks like. There are always very interesting behaviour patterns in each competition. One of them is the \"ensemble swarm\" once a few high-scoring kernels appeared. And another one is also the shake-up. ;-) Don't trust the public LB, start focussing on your CV and your comparison with the public LB, try to balance! \n\nThere is nothing bad on building ensembles as it has it's own goal to reduce overfitting. It's also not bad to built upon ideas that had been shared by the community. But instead of playing a \"point-and-click\" adventure of the random kind it's definitely better to built something you understand (at least to some point). \n\nWithout learning something during a competition all Kaggle medals are just an illusion of competence that can be easily dissolved when someone starts asking knowledge related questions.\n\nSo don't worry! :-) ", "1513849": "These notebooks do help people like me who are just starting out to learn and get better but yeah there should be like a right time to share these. Can't really see what the people who just copy their submissions are hoping to achieve. Like how long will you keep this up, the best will still beat you by the end and you really get nothing out of it.", "1510497": "I lost my interest in Kaggle because of the ensemble kernels shared.", "1514194": "This is why it is always hard to participate in comp as a newbie...", "1513742": "High scoring notebooks should not be allowed in last 3-4 weeks.", "1512379": "I found public notebooks (and specially those with high score) to be really informative about the good pratice in data science (here in deep learning). I don't think we should remove them because those are really informative and this is what make kaggle interesting.\n\nBut as the author say it is really frustrating to see people just copied / paste public submission. It makes the competition frustrating. As other comment put it, we should just stop public submission that are too high in the last days of the competition (~2 weeks before the deadline).", "1507835": "while I agree these notebooks are stupid/annoying they do tell you a little about who is merging in public notebooks into their ensemble (you see some teams dramatically shift when a public-no-real-info notebook is released)", "1511169": "Let us stand on the shoulders of giants😛", "1509173": "Cant believe there are dummy people copying others code. You are learning nothing. And for that Kaggle is, to learn. if you want money get a f****** job.", "1515112": "Is there a dislike button on kaggle?)", "1510871": "According me there must not be any public notebook for active competition.... Discussion is okay... 🧐🧐", "1559742": "Hey All,\n\nThank you all for taking part in our competition. The participation has been overwhelmingly positive. We are currently conducting a survey to gauge the demographic and outreach achieved. Kindly spare 2min and fill in this survey https://forms.gle/QP9L16niPexozyhu5.\n\nThank you all,\n\nRegards,\nChris", "1515401": "Kaggle is fun", "1515262": "Interesting", "1512518": "Hear Hear !", "1511060": "I also find it really hard to participate as an individual" } }